Am4 prozessor test: CPU Benchmarks Hierarchy 2022: Processor Ranking Charts

Ryzen 7 5800X3D review: Showing how far AMD’s Ryzen has come

AMD’s AM4 socket has been around five years now, and what a run it’s had. Since its launch alongside first-generation Ryzen CPUs, this socket has been a mainstay, with unprecedented support from Team Red that enables use of first-gen motherboards with current processors.

AM4 will soon cede center stage to AM5 later this year, but not everyone will need PCIe 5.0 or want to shell out for expensive DDR5 memory. That means Ryzen 5000 chips—the final batch of AM4 processors—could prove to be a much more cost-effective upgrade for people who want better performance.

That makes the recent launch of the Ryzen 7 5800X3D, AMD’s first CPU to boast its impressive new 3D V-Cache technology, a particularly tantalizing prospect. So when our sample crossed our desks, we decided to answer the question that many people would have. We cracked open our old Ryzen 7 1800X build, pulled the newer 3800X and 5800X off the shelf, and put them head to head with the 5800X3D, all to see just how much this new chip improves upon its predecessors—and to see how far AMD’s venerable AM4 has come since Ryzen’s early days.

Ryzen 7 5800X3D

How we tested

Our 2018 “Viva Las Vegas” build (shown here with four different lighting schemes) was repurposed for these benchmark comparisons.

Adam Patrick Murray / IDG

For our benchmarks, we repurposed and upgraded this Ryzen 7 1800X build from 2018. Its parts list is as follows:

  • CPU Cooler: Cooler Master MasterLiquid ML240L RGB
  • Motherboard: Asus ROG Crosshair VI Extreme X370
  • RAM: Corsair Dominator DDR4/3600 CL16 (2 x 16GB)
  • SSD: Corsair MP600 Pro XT PCIe 4.0 SSD
  • GPU: Nvidia GeForce RTX 3090 Founders Edition (driver ver. 512.15)
  • PSU:  Corsair RM750i 750W 80+ Gold
  • Case: Cooler Master H500P White Mesh
  • OS: Windows 10 21h3

The Asus X370 motherboard was updated to BIOS 8503 (Agesa v2 PI 1206), with PCIe Resizable BAR (a.k.a AMD’s Smart Access Memory) turned on. Core Performance Boost was on for the Ryzen 7 1800X, and Precision Boost Overdrive (PBO) was left on Auto for the 3800X, 5800X, and 5800X3D. The Cooler Master AIO cooler had its fans and pump set to 100 percent. Memory was clocked down to DDR4/2933 CL20 for full compatibility across all CPU generations. Windows 10 used the Best Performance power mode, with VBS and GPU hardware accelerated scheduling turned off. No encryption was enabled.

Our CPU line-up encompassed the Ryzen 7 1800X, 3800X, 5800X, and 5800X3D. Second-generation Ryzen chips weren’t included due to a lack of comparable part (AMD never produced a Ryzen 7 2800X) and a relatively mild difference between first- and second-gen Ryzen. We used the same build for all benchmarks, swapping the CPUs in and out of the motherboard to compile our set of benchmark results.

Specs for the CPUs used in our tests.

PCWorld

Performance results

As made clear when the Ryzen 7 5800X3D’s reviews published in late April, this processor has a clear, singularly defining purpose—gaming. (Not up to speed on the details? Here are 5 things you need to know about 5800X3D.) Compared to the standard Ryzen 7 5800X and contemporary Intel rivals like the Core i7-12700K and Core i9-12900K, it has a tangible impact at 1080p. It also offers smoother performance at higher resolutions like 1440p and 4K, with higher frame rates when looking at the worst one percent of performance (aka one-percent lows). However, The Ryzen 7 5800X3D is also weaker in productivity and content creation benchmarks thanks to clock speeds, which are slightly lower than the non-3D version.

That’s when talking about its contemporary rivals, though—the standard 5800X and Intel’s Core i7-12700K and Core i9-12900K. For these benchmarks, we’re examining the boost to performance you get compared to a Ryzen 7 1800X or 3800X, because it makes more sense to think of the 5800X3D as an upgrade chip, rather than the go-to for a new build.

And as you’ll see, ditching an older Ryzen CPU can result in mammoth performance gains, and not just in gaming.

Our video retrospective of AM4 dives deeper into these performance benchmarks.

Production

Longer bars indicate better performance (except for Blender, which is the opposite). You can click and drag each slide with your mouse (or swipe) to navigate between slides.

We’ll start with the production-oriented benchmarks, which are the 5800X3D’s weak spot compared to contemporary rivals—including its sibling, the standard 5800X. Just as in other 5800X3D reviews, this chip trailed behind the 5800X and its faster clock speeds in our rendering, video editing, and photo editing benchmarks.

But the slugfest between today’s chips isn’t the story we’re after here. Instead, we’re looking at how much performance you gain if you’ve got an older Ryzen chip. The answer is quite a dang lot if you’re a Ryzen 7 1800X owner. In test after test, you’re looking at sizable double-digit gains—between 29 and 60 percent for rendering, 27 percent in Premiere Pro, 50 percent for Photoshop and Lightroom.

For 3800X owners, an upgrade to the 5800X3D offers more modest gains—the biggest increases top out around the 15 percent mark. That’s not bad, but for most people who focus on this kind of use, not enough to compel a switch. A different, more affordable Ryzen 5000 series chip will be a better option.

Office productivity

Longer bars indicate better performance in Procyon 2.0. For the Excel eSports spreadsheet, it’s the opposite. You can click and drag each slide with your mouse (or swipe) to navigate between slides.

Most people don’t think much about performance in apps like Microsoft Word, but your processor can make a difference when creating slideshows, editing documents, or replying to email. And that’s what the aggregated scores for Procyon Office revealed. This benchmark, which measures performance in Word, Excel, PowerPoint, and Outlook, showed a whopping 57 percent improvement stepping up from an 1800X to a 5800X3D. 

That said, those who push Excel to the max won’t see as much of a gain, as shown in the benchmark created for us by Excel eSports competitor Jason Moore. But the 5800X3D was never the right chip for that kind of core-hungry work—as just seen above in the production benchmark results. Also similar to the production benchmarks, going from a 3800X to a 5800X3D offers a far more modest increase in performance. Procyon Office posts an 15 percent uptick in its aggregated score.

Would you upgrade a chip solely for Microsoft Office? Possibly, but the $450 5800X3D wouldn’t be the best choice for that. But these results certainly contribute to enhancing the chip’s appeal.

Web browsing

Longer bars indicate better performance. You can click and drag each slide with your mouse (or swipe) to navigate between slides.

Speaking of benchmarks that further the 5800X3D’s appeal—web browsing benchmarks aren’t the numbers that turn heads, but this kind of computer use is the practical stuff no one talks about but affects everyone. You’re going to use your PC to browse the web for at least some portion of your day.

Upgrading from an 1800X to a 5800X3D will make your time spent in Chrome, Edge, or Firefox all but sing. The results in Speedometer, Jetstream, and WebXPRT 4 show performance increases ranging from about 50 to 60 percent. Switching from a 3800X to a 5800X3D isn’t as dramatic, with gains between 13 and 22 percent. 

Again, you won’t be buying this chip to surf the web (well, perhaps you are, but most of us aren’t rolling in that kind of dough). But it’s nice to see that despite the lower performance in production tasks, the 5800X3D offers tangible benefits in other, more common types of PC use.

Gaming

Longer bars indicate better performance. More results can be found below. You can click and drag each slide with your mouse (or swipe) to navigate between slides.

You probably already guessed—the 5800X3D slays the 1800X in gaming, too. But the older 3800X sees more sizable gains in this area as well.

When looking at the average of the 17 games we benchmarked, the 5800X3D offers a staggering 72 percent performance increase compared to the 1800X at 1080p. For the 3800X, you’ll see a 38 percent bump. This outcome isn’t a total surprise if you’ve already read up on the 5800X3D—it clearly has an edge in gaming, even when pitted against contemporaries. (Note how it outdoes the standard 5800X by nine percent in those 17 games.)

You don’t lose this uptick immediately as you scale up in resolution, too. At 1440p, the 5800X3D still notably improved on the 1800X. In Shadow of the Tomb Raider, it was a 40 percent gain, while in Tom Clancy’s Rainbow Six Siege it was 101 percent. For the 3800X, it was 15 percent for Shadow of the Tomb Raider and nearly 40 percent for Rainbow Six Siege.

Longer bars indicate better performance. You can click and drag each slide with your mouse (or swipe) to navigate between slides.

Once you become GPU bound, the 1800X does begin to hold its own in some games. Moving up to a 5800X3D at 4K resolution is a mere two percent improvement in Shadow of the Tomb Raider. (For the 3800X, it’s a scant one percent.)

But as you can tell from the discussion of individual benchmarks, results can vary, and wildly at that. Switching from an 1800X to a 5800X3D in Metro Exodus at 1080p only nets a relatively mild improvement of 13 percent. But in Rainbow Six Siege at 4K? 26 percent.

So while replacing an 1800X with a 5800X3D will supercharge your 1080p gaming on average, gamers who play just a few games (or the same game regularly) should pay attention to individual benchmark results. That’s especially so if you play more often at higher resolutions.

Power draw

You can click and drag each slide with your mouse (or swipe) to navigate between slides.

We’ll take a quick look at power draw before wrapping up these benchmark results. This information is important if you care about how much electricity your system consumes and/or heat generated. The former matters if you live somewhere with expensive rates, and the latter is relevant if you live in a warm climate or have a PC case with limited cooling.

For this test, we ran Cinebench R23 on its default setting, which is a 10 minute loop of the render. Of the chips in our lineup, the 5800X3D and 3800X drew a bit less power than the 1800X, which still came in under the 5800X’s demands. The 5800X3D did have a few more spikes than the 3800X in the single-core benchmark, but not many overall.

Final thoughts

The two CPUs that bookend the AM4 era: the Ryzen 7 1800X (left) and the Ryzen 7 5800X3D (right).

Willis Lai / IDG

As we just saw, Ryzen 5000 smokes first-generation Zen. It posts a decent boost over Zen 2 as well. But the 5800X3D doesn’t settle the question of whether or not to upgrade on AM4. In fact, it actually complicates things a little more.

The Ryzen 7 1800X may seem pathetic based on some of the benchmark results above, but it still faithfully serves in a number of PCs, including the one still used by PCWorld’s very own Brad Chacos. Most owners have only been curious about an upgrade—not desperate. They’ve wondered about switching to Intel, waiting for Ryzen 7000, and if they have a compatible motherboard, going with a Ryzen 5000 chip.

The 5800X3D and its $450 price tag adds one more fork to the Ryzen 5000 decision tree, rather than being a solution that overwrites the existing 5600X, 5800X, and 5900X. You have to really think about your budget and how you use your PC. Performance isn’t the issue. Rather, if the point of dropping in a new AM4 CPU into your system is to boost performance while saving cash, another Ryzen 5000 chip could offer more bang for your buck. A 5600X ($200 street price) could make more sense for a 4K gamer, while a 5900X ($390 street price) offers more firepower for content creators.

The street price for the 12-core, 24-thread Ryzen 9 5900X is currently $390—or $60 less than the 5800X3D, which just launched and thus still goes for MSRP.

Gordon Mah Ung / IDG

That said, choosing the 5800X3D is still more cost-effective than replacing your processor, motherboard, and possibly also memory—as you’d have to do with an upgrade involving an Intel processor or Ryzen 7000. Moreover, the fact that you even have these granular choices is incredible. AM4 has lasted four CPU generations with full backward compatibility. Meanwhile, the competition can’t always be relied upon to last two generations. To go from a Ryzen 7 1800X and its early democratization of core count to a Ryzen 7 5800X3D and its face-melting performance in games is a simple CPU replacement. That’s it. You don’t have to swap anything immediately, either—you’ll probably be able to snag the 5800X3D for less once Ryzen 7000 launches later this year. PC building was not like this before.

Speaking of Ryzen 7000, AM5 will be around for a while, too. During a CES 2022 interview, AMD CEO Dr. Lisa Su said that as with AM4, we should expect AM5 to be a long-lived platform as well. No exact number of years were given, but with the precedent set, here’s to hoping for a continued legacy of generous support.

Editor’s note: This article was updated on 6/10/2022 to include our YouTube retrospective video.

AMD Ryzen 3 1300X And Ryzen 3 1200 AM4 CPU Review — Page 2 Of 2

« AMD Ryzen 3 1300X and Ryzen 3 1200 AM4  CPU

Test System and Benchmarks

Processor AMD R3 1300X, R3 1200, R5 1600X, R5 1500X, FX-6350, FX-8350, Intel Core i3-6100
CPU Cooler
AMD Wraith Stealth (Ryzen 3), Stock Intel (i3-6100), Wraith Max (FX and Ryzen 5)
Power Supply
Corsair HX850W 80 Plus Gold PSU
Memory Corsair Vengeance LP 2x8GB @ 2666MHz CL16, Mushkin Stealth 2x4GB DDR3 @ 2133MHz CL11 (FX-8350)
Storage OCZ Agility 4 256GB SSD (OS)
Graphics Gigabyte GTX 960 Windforce 3x OC
Motherboard(s) ASRock AB350 Gaming K4 (3. 0), Aorus Z270X-Gaming 5 (F8), MSI B250M-Mortar, ASUS M5A99FX EVO R2.0 (2501)
Operating System Windows 10 Pro x64 (1704)

 

Price

The new Ryzen 3 directly replaces the FX-6350 and FX-8350 within the same price point, the latter two priced with their bundled Wraith cooler as well. They are still available for $25 less without the cooler. The i3-6100 is the previous generation Skylake i3. Although older than the current generation Kaby Lake i3-7100, it is still similarly priced (since Intel does not really drop prices even when the new generation comes out) and should still pose quite the challenge to the Ryzen 3’s, especially in single thread applications.

CPU Cores/Threads Clock Speed (Turbo) L3 TDP
Intel i3-6100 2/4 3.7 GHz 3MB 51W
AMD FX-6350 6/6 3. 9 (4.2) GHz 8MB 125W
AMD FX-8350 8/8 4.0 (4.2) GHz 8MB 125W
AMD Ryzen 5 1600X 6/12 3.6 (4.0) + 100 XFR 16MB 95W
AMD Ryzen 5 1500X 4/8 3.2 (3.7) + 200 XFR 16MB 65W
AMD Ryzen 3 1300X 4/4 3.5 (3.7) + 200 XFR 8MB 65W
AMD Ryzen 3 1200 4/4 3.1 (3.4) + 50 XFR 8MB 65W

For those running an i3-6100 or on an FX platform, these benchmarks should give an overview how much advantage moving to the new platform provides and if it offers considerable value.

CPU and Memory Benchmarks

The AIDA64 suite has various benchmarks for CPU, FPU and memory testing: CPU Queen is an integer benchmark that tests branch prediction and misprediction penalties. CPU PhotoWorxx tests the SIMD integer arithmetic execution units of the CPU and the memory subsystem. CPU ZLib is a compression benchmark that tests the combined CPU and memory performance. CPU AES is a multi-core encryption benchmark that uses Advanced Encryption Standard data encryption. CPU Hash is an integer benchmark that measures performance using SHA1 hashing algorithm.

Advertisement

Using AIDA64 as well, memory and cache performance is benchmarked as well as FPU. FPU Mandel measures double precision FP, FPU SinJulia measures extended precision floating point.

MAXON’s Cinebench R15 runs two sets of benchmarks, a processor test for the CPU and an OpenGL test for the graphics processor. The 3D rendering workload is based on Maxon’s Cinema 4D, a program utilized by movie production studios world wide for special effects so it is a semi-synthetic test closer to real-world 3D graphics benchmarking.

Multimedia Benchmarks

x264 is a popular free software library for encoding video streams into the H.264/MPEG-4 AVC format. x264 FHD measures how efficient a system is in encoding H.264 video and produces results in frames-per-second. H.265/HEVC video encoding is the future of video able to compress significantly larger resolution videos including 4K and make streaming feasible. The downside is that hardware support is still few and far between so processing requirement is steeper than current H.264/AVC standards. x.265 is an open-source implementation of the H.265 standard and x.265 HD benchmark tests the CPU’s ability to process an HEVC video.

Handbrake is free open-source video transcoder that can convert various video file formats to compatible video files for other typical media viewing applications such as in an iPhone or Android tablet. A 52:55 1.1GB H.264 MKV file was used converted to the built-in iPod Preset (5G Support), then manually timed. The results are in seconds and the lower number is the better result.

Advertisement

The Ryzen CPUs shine on multimedia tasks.

Compression Benchmarks

7-zip is an open source (GNU) compression program utilizing LZMA method as the default. WinRAR on the other hand is a Windows version of a popular compression software created by Eugene Roshal in 1993, widely used for its flexibility. Both have built in benchmarks that measure the system’s compression and decompression capability.

Having a larger core count advantage is obvious here for the FX and the Ryzen 5 CPUs.

Gaming Benchmarks

Futuremark’s 3DMark is a semi-synthetic gaming benchmark that calculates both graphics and CPU-bound physics in a controlled series of tests and provides scores that can be compared with other gaming platforms. Unigine Heaven is a synthetic benchmark that is completely GPU bound for testing possible PCI-E graphics performance inconsistencies.

It is interesting to see the aging FX CPUs still clinging on on the synthetic Physics score and holds well against the Ryzen 3’s. For the purely GPU-centric loads such as Unigine Heaven there is virtually no difference detected in performance, but the lower min-frames are very noticeable on the older CPUs with real-life games.

Temperature and Power Consumption:

AMD has “fixed” the temperature readings of Ryzen with the latest Ryzen Master software, now measuring the true junction temperature. Tested with the silent Wraith Stealth fan included, the temperatures barely reached the idle readings from the pre-update software. Although the Wraith Stealth is not as capable as the larger Wraith MAX RGB cooler used on the R7 and R5 CPUs in this graph, it still does a decent job of keeping temperatures in check.

Overclocking

Overclocking is a mixed bag with Ryzens. While it is not thermally limited, Ryzen CPUs including the Ryzen 3s cap out at around 3.8 to 4GHz. The advantage is that unlike Intel’s i3s, all Ryzen CPUs are fully unlocked on any B350 or X370 motherboard. More importantly, the memory can be pushed further than what is standard JEDEC spec. Paired with the latest AGESA 1.0.0.6 BIOS or newer, users can expect  performance headroom up to 3200MHz DDR4 or higher.

Intel’s B250 motherboards however, cannot adjust the memory beyond the 2400MHz limitation imposed by Intel. One major advantage that Intel has over the Ryzens is that strong single-threaded performance. That should be obvious considering how much of a headstart they had, but the good news is that AMD has quickly closed that gap and it appears they will continue to do so for the foreseeable future under the leadership of CEO Dr. Lisa Su.

The Wraith Stealth coolers do a decent job keeping temperatures in check but it has to be cranked up to max fan speed when overclocking. The Ryzen 3 1300X even at stock is already a lot of work for the Wraith Stealth on silent fan speed preset that it has to be set to Standard or full-speed for truly comfortable temperatures.

Final Thoughts:

I don’t need to tell you that AMD is back in a big way. The Ryzen 7 and Ryzen 5 are extremely impressive and they even have an HEDT offering now with the Ryzen Threadripper. The Ryzen 3 follows suit, showing a significant leap in performance compared to previous generation FX and Athlon CPUs. It is also priced competitively, undercutting Intel’s i3 offerings.

The real strength of the Ryzen 3 processors, just like the Ryzen 5 is in combination with the B350 platform. This provides users with a very affordable and more importantly, upgradeable path. This is something that is ingrained in AMD’s identity as a company and it is nice to see that they have not abandoned that value-for-performance approach.

There is also the value addition of the Wraith Stealth coolers which should not be ignored. AMD hit the ball out of the park with their Wraith and Wraith MAX but they also did well with the Wraith Stealth since it is perfect for Ryzen 3’s. It operates silently in PWM mode and the installation is very sturdy, using spring-loaded screws instead of flimsy push-pins that Intel used before ditching the heatsink entirely. It also helps that the Ryzen 3, like the Ryzen 5 and 7 are very efficient and thermally optimal that it does not require a considerable cooling solution to tame, or worse, require delidding.

« AMD Ryzen 3 1300X and Ryzen 3 1200 AM4  CPU

Previous page 1 2

Related Articles

Test and review: AMD Ryzen 7 5800X3D is a great upgrade for socket AM4

A little over a week ago, the first tests of the Ryzen 7 5800X3D, a processor with a 3D V-cache cache that is aimed at gamers, appeared on the Web. Meanwhile, the CPU began to appear in European retail. The Ryzen 7 5800X3D has made it to our test lab, so we’re excited to present our own benchmarks. The processor is also interesting in that it is AMD’s latest blow for Socket AM4, since the transition to AM5 should be expected in the second half of the year. The Ryzen 7 5800X3D will defend AMD’s position in the gaming segment until the new Zen 4 models arrive. Interestingly, AMD plans to expand its 3D-V cache strategy with the new generation, although the benefits are not visible in all scenarios. nine0004

AMD announced the first Ryzen processors with 3D V-cache back at Computex 2021. But then it was not entirely clear how AMD would use the new technology. The first CPUs were supposed to appear at the end of 2021, but they had to wait until April 20, 2022. On this day, sales of the Ryzen 7 5800X3D, the first CPU with a new cache, started, that is, many months after the official introduction. Such a long wait time led to criticism from gamers and AMD fans, yet the processor arrived quite late. And today it does not look as impressive as at the end of 2021. nine0005

For the first generation of Ryzen with 3D V-Cache, AMD opted for processors with only one CCD die, meaning eight cores. On the other hand, they are quite enough for games. Last year, 3D V-cache technology was introduced with a model containing 12 cores on two CCDs. Apparently, problems with the availability of crystals affected, nevertheless, EPYC processors with an increased cache can be sold for much more. In the AMD range, the Ryzen 7 5800X3D processor is positioned as follows. nine0005

Subscribe to the Hardwareluxx VKontakte and Facebook groups, as well as to our Telegram channel (@hardwareluxxrussia).

nine0018 3.7 / 4.8 GHz

Ryzen 5000 line
Cores Base frequency/Boost L3 cache TDP Price
Ryzen 9 5900X 12 64 MB 105 W 45.200 ₽
Ryzen 7 5800X3D 8 3.4 / 4.5 GHz 96 MB 105 W EUR 489
Ryzen 7 5800X 8 3.8 / 4.7 GHz 32 MB 105 W 32.100 ₽

Like the regular 5800X, the Ryzen 7 5800X3D has eight Zen 3 cores, but the base frequency and Boost are still lower. The TDP of 105W hasn’t changed, so the additional cache’s power consumption was probably the reason why AMD couldn’t set identical clock speeds. The 32MB of L3 cache in the Ryzen 7 5800X CCD of the Ryzen 7 5800X3D is complemented by 64MB of 3D V-cache, for a total capacity of 96MB. Therefore, the processor has even more L3 cache than a CPU with two CCD crystals (like the same Ryzen 95900X).

Otherwise, no difference. The Ryzen 7 5800X3D works with DDR4 memory in two channels and provides support for PCI Express 4.0. The only unique advantage of the processor is the additional cache.

AMD has set the price tag at €489, which is significantly higher than the Ryzen 7 5800X and even more expensive than the 12-core model. Of the direct competitors of the Ryzen 7 5800X3D, we can note the Core i7 12700K (test) for 56,000 ₽ and the Core i9-12900K (test) for 71,800 ₽. The Core i9-12900KS processor (test), which has not yet appeared in Russia, costs significantly more. Ultimately, let’s see how the processors show themselves in terms of price / performance ratio. nine0005

3D V-Cache in detail

Here we would like to remind you once again how the 3D V-Cache technology works. Enclosure technologies became the starting point for the development of a new cache. Whether it’s HBM in the GPU package or the shift to chiplets in the CPU segment, AMD can make a claim in these technologies. On the other hand, AMD is still heavily dependent on partners. It was only in collaboration with partners that AMD was able to develop a product in this form and bring it to market. nine0005

3D stacking technology is still the final stage of AMD development. But putting a layer of SRAM on the processor die is only the first step. In the future, we can expect multi-layer SRAM chips, which can significantly increase the cache capacity.

The 3D V-cache is connected to the CCD directly via TSV (through-silicon via), which was made possible thanks to the 3D packaging technology that AMD developed together with TSMC. The CCD, which is manufactured using a 7nm process, is rotated 180° and 95% of the crystal is sawn off. Of the thickness of 400 microns, only 20 microns remain. And just above this thin layer is the 3D V-Cache cache. It is also produced using a 7nm process technology, the crystal area is 41 mm².

The connection between CCD and SRAM is made with copper TSVs. As AMD points out, the adhesion of the TSV copper pads on both sides is sufficient to connect and hold the two dies. The connection bandwidth is more than 2 TB / s. An extra layer called «structural silicon» (D1A and D1B) adds height where the 3D V-Cache doesn’t, so the thickness of the CCD stays the same. nine0005

Comparison of process technology, area and number of transistors
Process Chip area Number of transistors
CCD 7 nm 80. 7 mm² 4.15 billion
IOD 12 nm 125 mm² 2.09 billion
L3D 7 nm 41 mm² 4.7 billion

In the table above you can see the number of transistors and area in relation to other chipsets. The CCD with eight Zen 3 cores and 32 MB of L3 cache has an area of ​​80.7 mm² and 4.15 billion transistors. The IOD adds 125 mm² and 2.09 billion transistors to the package. IOD and two CCDs can already be called a rather complex structure in Ryzen and without 3D V-cache. If you add the L3D cache, then about half of the 80 mm² CCD die will be covered by the additional cache. You should also add 4.7 billion transistors to the final number. And here it means not only SRAM, but also control logic, power supply and other components. nine0005

There are no questions about the structure of the CCD with the L3D cache. But there seems to be a significant difference between diagrams and practical implementation.

According to information materials, it seems that SRAM is located on the CCD just in the center, above the L3 cache area. Layers of «Structural silicone» are added on the left and right to compensate for the difference in thickness. But after the ISSCC 2022 presentation, it became known that there is another layer of Support Silicon above them. This layer ensures that the thickness of the CCD is exactly the same as the CCD without 3D V-cache, so as not to change the physical characteristics of the package. nine0005

The image above shows a scalped AMD EPYC 7473X processor, which was then further analyzed for its die structure. The photo on the right shows one CCD die with the top layer of Support Silicon. It most likely aids in transferring heat to the Faucet, although CCD without 3D V-cache is the same thickness with passive silicon layers up to the Faucet. The pictures are from user @wassickt, who does chip analysis quite professionally.

We recommend checking out our guide to picking the best Intel and AMD processor for the current quarter. It will help you choose the best CPU for your money and not get confused in the range of models on the market. nine0005

<>Test and review: AMD Ryzen 7 5800X3D is a great upgrade for socket AM4
Test configuration

What to upgrade AMD Ryzen 5 1400 to? Ryzen 5 5600X vs Ryzen 7 5800X3D — i2HARD

test & comparison So we found the Ryzen 5 1400 processors lying around at the bottom of the box. And immediately when we saw it, the thought came: “A new socket is coming, it’s time to evaluate the power of the upgrade, undoubtedly, the long-playing AM4.” And then the ideas poured in. What can a person with such a processor upgrade to now? Ryzen 7 5800X3D is the best gaming solution from the reds on AM4, we will definitely take it for comparison in order to calculate the growth factor. But what if the person doesn’t want to take an expensive processor that is a dead end for that socket? Then let’s add another 5600X to the comparison! An excellent price / performance ratio will not make you regret the purchase, and to make it even more interesting, we will compare it in several stages with and without an emphasis on a video card. And as a bonus, we will have 5800X3D in full stuffing. The owner of the R5 1400 is unlikely to have bought a dropsy, a top-end motherboard and memory at one time, so he may wonder if his components are too weak for such powerful hardware. nine0005

Contents:

  • Test bench
  • Stock tests
  • Synthetic tests
  • Temperatures and consumption
  • Tests in games
  • Overclocking tests
  • Undervolt, processor and memory setup
  • Synthetic tests
  • Temperatures and consumption
  • Tests in games
  • Conclusion

Test bench

  • Video card #1: INNO3D GeForce RTX 3050 TWIN X2 OC
  • Graphics Card #2: Palit GeForce RTX 3080 Ti GameRock OC
  • Processor #1: AMD Ryzen 5 1400
  • Processor #2: AMD Ryzen 5 5600X
  • Processor #3: AMD Ryzen 7 5800X3D
  • MSI B450M Bazooka MAX WIFI
  • Motherboard #2: ASRock B550 Taichi Razer Edition
  • RAM #1: Crucial Ballistix Sport LT [BLS8G4D30AESCK] 2x8GB
  • RAM #2: G. SKILL Trident Z F4-3200G14 2×16 GB
  • Cooling System #1: ID-Cooling SE-224-XT BLACK V2
  • Cooling System #2: ARCTIC Cooling Liquid Freezer II-360
  • Storage: Crucial MX500 2 TB
  • Power Supply: Deepcool PQ850M ATX 850W Gold
  • Chassis: BC1 Open Benchtable
  • Operating system: Windows 10

It turns out that in the test bench we will have the above-mentioned processors installed in the MSI B450M Bazooka MAX motherboard and cooled by the usual ID-Cooling SE-224-XT bag. In addition, the kit includes two 8-GB Crucial Ballistix modules on micron rev.E chips (rev.I). nine0005

The most equipped 3D variant will be surrounded by the Arctic Liquid Freezer II-360 LSS, two 16 GB modules based on Samsung B-Die chips and an ASRock B550 Taichi Razer Edition motherboard.

Two video cards. RTX 3050 and 3080 Ti.

Stock tests

Synthetic tests

So, the first stage. Having bought an RTX 3050, 2060 or something on the level for his ryzen 5 1400, our imaginary user realized that the processor would be rather weak. Should he fork out for a new one, and which one? nine0005

In AIDA64, the cache speed of 5800X3D with different mother/cooler/memory kits is practically indistinguishable, but higher frequency XMP with better timings gives about 10% additional bandwidth and reduces latency by almost 6%.

If the user often uses working software that uses mainly the processor, the increase will not be weak. The 5600X is 81% faster in single thread alone, and 1.5 times more cores results in a 2.5x increase in multithread. The 5800X3D is not the best choice for this kind of task for the money, but 8 cores still make themselves felt. The high-performance cooling system and motherboard allow it to handle higher frequencies, which means you get an additional 1% in single-threaded performance and 3% in multi-threaded performance. nine0005

In Cinebench R23, the gap between new ryzens and 1400 increased. The 5800X3D in a more budget board even scored a little more points in a single thread than with more expensive components. It looks like MSI boost is more aggressive. However, the 5600X beats it with a higher frequency.

In Geekbench 5, the memory subsystem does its bit, so we see a 9% advantage for the kit with the faster XMP.

In Premiere Pro, rendering a simple project on the processor will save half the time when upgrading to a 5600X. Additional payment for 5800X3D will give another 17% acceleration, and the combination of all factors in a more expensive kit leads to a 14% increase. Here, faster memory helps a lot, and PCI Express 4.0 instead of 3.0 could have an effect. nine0005

Temperatures and consumption

Well, temperatures, where without them. With a liquid cooling system, the processor warmed up by 10-12 ° C weaker. Four- and six-core processors look like cold processors against the background.

In a more serious stress test, even underwater, the 5800X3D hits a temperature limit of 90°C. With it, the processor took advantage of the thermal margin and due to this it works at a higher voltage, and, accordingly, frequencies. Whether 170 MHz is worth it is an open question, given that this is only 4% against the background of the total frequency. nine0005

An equally important issue is the motherboard VRM temperature. 8 cores are 8 cores. But the 5800X3D consumes less power than the regular version, so the range of suitable motherboards is much larger. With an average consumption of 115 W, the motherboard heats up within reasonable limits. In which case, we remember the simplest collective farm, consisting of one piece of paper, but capable of helping a lot. Well, let’s not forget about the undervolt. We will also talk about him.

Tests in games

Synthetics are synthetics, and 5800X3D is more designed for games, so the emphasis will be on them.

Cyberpunk, medium graphics preset, low crowd density, 1080p, SMT patch enabled. We remind you that we currently have an RTX 3050 installed in the system, unable to render 60 FPS even at medium settings. And people still ask “Why do we need DLSS in Full HD?”. It is clear that it is not difficult for the 5600X and 5800X3D to prepare so many frames for the 3050, but 1400 bogs down, sometimes dropping to 30 FPS. 8 GB of VRAM is more than enough for our settings, so PCIe 4.0 provides a subtle advantage over the 3rd version. nine0005

Far Cry 6, no tracing ultra preset, 1080p. Unlike Cyberpunk, here the graphics card feels much better, allowing you to play around with tracing. Ryzen 1400 also prepares noticeably more frames, and this despite the fact that it is only half loaded. Again, there is no difference between a six- and an eight-core processor, we are severely limited by 3050. PCI Express 4.0 gives a 4% increase similar to the previous one.

Shadow of Lara, highest preset, 1080p. An even less demanding game for the video card does not allow us to see the difference between older processors, because the requirements for them have also decreased. This is especially evident in the 1400 ryzen, which only drops below 60 FPS on the market, but the statistics of rare and very rare events are spoiled by strong lags during attempts to load the mountain. Not only PCI Express makes itself felt when focusing on the video card on a more expensive system. In addition to it, we see a higher boost on the 5800X3D with a liquid cooling system. Thanks to it, the temperature is lower by at least 10 ° C. nine0005

CS:GO, minimum settings, maximum anti-aliasing. It was unexpected to see in this game the highest impact of PCI Express bandwidth. The 4th version gave us a 10% increase. However, it is extremely problematic to feel it, given the available FPS. Even the ryzen 1400 renders over 150fps.

Summary. If you have an RTX 3050 in the system for a long time, and the computer is aimed at games, then paying extra for the 5800X3D is not worth it. It costs 3 times more and the FPS is the same. For the 5600X, the cooling system and motherboard requirements are low, so you don’t have to worry about them. Remains PCI Express. Yes, we saw an increase even with a sufficient amount of video memory, but it’s not worth changing the motherboard for it, it’s too small. nine0005

If there are a lot of synthetics in life, then it is better to look at the 5700X, which is much cheaper, or at the twelve- and sixteen-core, close in budget to the 3D-shka.

Well, what if you plan to replace the video card with something like 3080 Ti? Does it make sense then to pay extra or again everything will rest against the card? Let’s go check.

We will leave the settings the same so that we can compare the results, but we will increase the resolution to Quad HD, which is more logical for the owner of such a video card.

In Cyberpunk, the 5600X complements the 3080 Ti almost perfectly, a little slower in preparing shots for it, only 4.5% was not enough, judging by the 3D. PCIe 4th version gave exactly the same 4% increase as with 3050. Without monitoring, you won’t even notice. But what you definitely can’t miss is more than a twofold increase compared to 1400.

In Far Cry 6, walking down a city street makes the CPU sweat. Well, to sweat, the 5600X is not even half loaded, because the game parallels as well as 5 parts ago. Nevertheless, the impressive L3 cache makes itself felt, the emphasis on the 5800X3D can only be seen at the end of the benchmark, when the view of the whole street drops it to 125 frames per second, while the 5600X produces about 90.

In Lara, the 5600X again falls short of the 3080Ti. In the market, the load on the processor and memory subsystem is seriously increasing, due to which the 3D-scale is a little ahead. Yes, for this game FPS can be called redundant, but we are considering situations in a general sense, because there is more than one game that has similar behavior.

Compared to 3050, the FPS has doubled, so the 5800X3D has a heavy burden on its shoulders. You can see it in temperatures. If with a budget cooling system the temperature was around 60 ° C, now it has risen to 71 ° C, especially since part of the 350 W emitted by the video card enters the cooler blades as hot air. The liquid cooling system is devoid of this disadvantage, on the contrary, it throws a little heat on the video card, but the gap from the air is still reduced from 11 to 9°C.

In Contra, there is no emphasis on the video card, only the processor performance is compared. The 5000 series processors are almost 3 times faster than the 1400 series, and the 5600X leads in average FPS due to the frequency, because its L3 cache is enough for CS: GO, but the 5800X3D is still ahead in rare and very rare events statistics.

With the 3080 Ti, the Ryzen 5 1400 was more than twice behind the 5600X, but the 5800X3D no longer showed any huge advantage. Why is that? It’s all about the emphasis on the video card. In our small selection of games, it was not only in Contra. Does this mean that if the user does not purchase a more powerful video card, then he does not need something more than 5600X or 5700X? No, it doesn’t. Because if you take the recently released Spider-Man, you can see the same picture as at the end of the Far Cry benchmark. At the highest graphics preset in Quad HD resolution, the 5600X renders around 94 FPS, 5800X3D — about 124. And this time, the 5600X is not balancing with the 3080 Ti, but an octa-core.

What are 94 and 124 FPS? This is a 32% difference. That is, in games where the 5600X will prepare 45 frames, with a 3D-shchika it will turn out 60.

Therefore, the next step is to lower the resolution to completely get rid of the emphasis on the video card, and take a closer look at the difference between these two processors.

In Cyberpunk with 5800X3D without focusing on GPU, FPS increased by more than 1.5 times, and compared to 5600X, the 3D cabinet is faster by an impressive 43%, and compared to the quad-core, FPS increased by 3.5 times! Also, fast memory and LSS began to make themselves felt, thanks to which the processor adheres to higher frequencies. The combination of these factors gave another 8% FPS. nine0005

Far Cry again shows more than 40% of the difference between 5600 and 5800, but the expensive «strapping» did not give much. Only 3%. There is a threefold difference between the extreme processors. And we have not yet compared this with a 1200 ryzen or an athlone!

In Lara, even with the maximum reduced resolution, the video card loading is sometimes higher than 90%. The expanded L3 cache of the 5800X3D opens up to its full potential, the average FPS without focusing on the GPU has increased from 182 to 289, the record 61% difference with the 5600X really looks very tempting, and with 1400 ryzen the increase is more than four times! The temperatures of the 3D kit with both kits rose another 10°C, maintaining the same gap. Higher core and memory frequencies gave a 4% increase. nine0005

Now 5800X3D doesn’t look like an empty overpayment. Is not it? For video cards, there is salvation in the form of upscalers, which is not for processors, so with future video cards there will be more and more such situations where the emphasis is on the processor. The last «but» remains. As we found out from the video dedicated to it, the 5800X3D gets a very weak boost from memory overclocking, because a large cache implies fewer RAM accesses. At the same time, the 5600X will have a larger gain, so, will it not turn out that after overclocking, this gap will be greatly reduced? And yes, we also promised to show the undervolt. nine0005

Overclocking tests

Undervolt, processor and memory tuning

Ryzen 1400 received a decent increase in cores. Their frequency is now fixed at 3.8 GHz, and the memory is overclocked to 3600 MHz with the first timing of 14.

The 5800X3D does not have a PBO setting in the BIOS. Instead, MSI suggests using Kombo Strike. In the maximum third mode, it is the same as setting the Curve optimizer to -30 across all cores. The voltage undervolt and LLC setting are also available, but even a slight decrease in them leads to a decrease in the effective frequency and, accordingly, a loss in performance. From memory, 3733 MHz came out with the first timing of 14.

Unlike the budget MSI based on the B450 chipset, the expensive ASRock Taichi has.. an empty BIOS. In addition to memory, there is nothing to configure here, even the Resizable Bar cannot be turned on. That’s the support. The frequency ceiling is the same — 3733 MHz, but B-Die allowed to tighten the timings much more strongly.

What to do if there is no way to configure the processor in the BIOS? This is where third-party software comes to the rescue, namely PBO2 Tuner. With this program, you can configure curva for all cores, just enter the desired value and click Apply. 3Ds tend to have a lot of headroom, which is why many work at the -30 minimum. nine0005

However, stability needs to be checked; in our case, CoreCycler was used, which tests each core separately. It is important during its operation to reduce the background load as much as possible. Then it remains to add the program to autorun. To do this, it is enough to create a shortcut on the desktop, in its properties to prescribe the necessary arguments, which are the same numbers that we entered in the application itself, so that when it starts, it will immediately apply our values ​​​​and close without an eyesore. It remains only to move this shortcut to the startup folder along the path C:\Users\*UserName*\AppData\Roaming\Microsoft\Windows\Start Menu\Programs\Startup. Lighter than easy. nine0005

The frequency of both memory kits in overclocking is the same, but the timings and the number of ranks differ. If with XMP the difference in throughput was about 10%, now it is full parity in terms of write speed, reading hardly differs, and only copying is 6% faster. By latency of 6%, the advantage was reduced to 4%.

Tuning the PBO algorithms, of course, helps to increase the frequency of the processor, but compared to manually overclocking the Ryzen 5 1400, the increase is small, because it increased its frequency by 19% for multi-threaded tasks. Especially in a single thread, it is noticeable that the once 75% advantage is reduced to 52%. And yes, now the result of an expensive and cheap set with an undervolt is the same.

Synthetic benchmarks

There is a slight difference in Cinebench, but barely noticeable. The 1400 improved its score by 15% in single stream and by 18% in multi stream. 5600X by 4% and 10% respectively, while 5800X3D accelerated only by 3% in multi-threaded workloads.

In GeekBench, overclocking ballistics also helped to seriously reduce 9% lag behind the expensive kit up to 3%.

Only in Premiere Pro, the difference between the Ryzen 1400, 5600X and 5800X3D within the same motherboard and other things has remained the same, the advantage of expensive components has again decreased to a barely noticeable level.

Temperatures and consumption

An interesting situation has arisen regarding temperatures. The 1400 became 11-12°C hotter, the 5600X without limits increased its temperature by 17°C, and the 5800X with the budget kit, on the contrary, became 11°C colder in the average temperature and 4 in the maximum temperature. With a liquid cooling system, it turned out differently. The average temperature dropped slightly — by 3 °C, and the maximum — by 6 °C. nine0005

In the air stress test, the 5800X3D still hits the temperature limit of 90°C, but now its frequencies are even slightly higher than it was with dropsy in the drain. Under water, the frequency increased not so much, but the temperatures dropped by 10 ° C. What’s funny, younger processors heat up more in Premiere Pro, and the 3D-scale, on the contrary, is in Hades.

Tests in games

Quad-core gameplay in Cyberpunk has become much better. Instead of 41 FPS on average, we get 55, which is a third higher. The new ryzens can’t boast that much growth. The 5600X got 22% and the 5800X3D got less than 8%. With an expensive kit, it’s 4%. As a result, the 3D lead over the 5600X has dropped from 43% to 27%, which is still decent, but not that impressive. nine0005

In Far Cry 6, the Ryzen 1400 drops below 60 FPS only at the end of the benchmark, which is a very good indicator. Overclocking gave it a little less this time, 29%, and this is clearly a good opportunity to hold out this processor a little longer. The 5600X received a 2 times smaller increase — 15%, and the eight-core under the tower and dropsy 4 and 3%, respectively. The difference between 5800 and 5600 is now 28% instead of 41.

In Lara, the mountain rendered much faster this time at 1400. His FPS increased by 30% again. The 5600X is up 17%, and the 3D box is up 3% and 1% with cheap and expensive strapping, respectively. From 62% superiority, there is still an impressive 44% advantage. nine0005

A low number of cache misses negated the gain from memory overclocking. The 3D-frame received a symbolic 1% in both cases. Even less. 5600X received a 5% increase due to the increased frequency, and 1400 with its 8 MB L3 cache does not change itself, it has become 28% faster.

Conclusion

Let’s summarize. Against the backdrop of modern processors, the Ryzen 1400 is undoubtedly weak, you already knew that without us, the 5600X is up to 2.5 times faster in games, and the 3D-scale is 3-4 times faster. But it is important to understand a few points that we found out. The 5800X3D does not require top components around it and is not so responsive to overclocking. Yes, they give an increase, but not so much that you are forced to do something. From all that could be noticed, 2 points stood out. Better XMP sped up renders in premiere by 14%, but does it make a difference if you wait 7 or 8 minutes, or even the difference reaches an hour in a multi-hour render. If this is important to you, then you should have chosen a twelve-core. Temperatures were the second. Under water, the 3D-scale often warmed up by 10 ° C less, yes, but under air it adhered to operating temperatures, they are simply unpleasant to the eye. Undervolt throws off the same 10 ° C, and it’s free, and in combination with overclocking the memory, the difference between these kits in games and synthetics has been reduced to negligible. PCI Express 4th version gave an increase everywhere, but so far there is enough video memory, both with 3050 and 3080 Ti, the difference is small. If you’re looking at a low-VRAM graphics card, then it might be worth considering. If the motherboard has a heatsink, then in most cases games and short-term loads will not cause even serious temperatures, not to mention throttling.