Home / Technology / An upgrade over regular threadripper?

An upgrade over regular threadripper?

Since the launch of AMD’s Threadripper Pro platform, the desire to see what eight memory channels bring to compute over the regular quad-channel Threadripper has been an exciting prospect. Threadripper Pro is effectively a faster version of AMD’s EPYC, limited to the use of a single CPU workstation, but also alerts a full 280 W TDP to match the frequencies of the standard Threadripper line. There is a 37% price premium from Threadripper to Threadripper Pro, which allows for ECC memory support, dual PCIe lanes and dual memory bandwidth. In this review, we compare each member of both platforms that are commercially available.

Threadripper Pro: Born of Need

When AMD began the journey with the new Ryzen portfolio, the definition of where each product sat in the traditional market has not always been completely clear. The first generation Ryzen was earmarked for standard users, but the top of the eight-core Ryzen 7 1

800X line competed with Intel’s high-end desktop market. The Zen 2-based portfolio saw mainstream Ryzen go to 16 cores, pushing past Intel’s best 18-core HEDT processor at the time in most tests. That the Zen 2-based Ryzen 9 3950X was still classified as a ‘mainstream platform’ processor, as it only had 24 PCIe lanes and dual-channel memory, sufficient for mainstream users, but not enough for workstation markets. These standard processors were also limited to 105 W TDP.

At the other end of the scale was the AMD EPYC, with the first generation EPYC 7601 with 32 cores, and the second generation EPYC 7742 with 64 cores, up to 225 W TDP. These share the same LGA4094 connector, have eight channels of memory, full ECC support and 128 PCIe lanes (first PCIe 3.0, then PCIe 4.0), with dual-socket support. For workstation users interested in EPYC, AMD launched single-socket P versions. These offered the same features, at around 200 TDP, and lost performance to the regular non-P versions.

AMD then launched Threadripper, a high-end desktop version of EPYC that went all the way up to 280 W for peak frequency and performance. Threadripper sat over Ryzen with 64 PCIe lanes and four-channel memory, so that regular users who wanted a little more, could get a little more. However, workstation users noted that while 280 W was good, it lacked official ECC memory support, and compared to EPYC, the reduced memory channel support and reduced PCIe compared to EPYC Threadripper stopped being adopted.

Then enter Threadripper Pro, which sits between Threadripper and EPYC, and in this case very much more on the EPYC page. Threadripper Pro has almost all the features of AMD’s EPYC platform, but in a 280W thermal envelope. It has eight channels of memory support, all 128 PCIe 4.0 lanes, and can support ECC. The only disadvantage of EPYC is that it can only be used in single-socket systems, and the top memory support is halved (from 4 TB to 2 TB). Threadripper Pro also has a small prize pool.

AMD comparison
AnandTech Ryzen Thread stopper Thread stopper
The company
Kernels 6-16 32-64 12-64 16-64
Architecture Zen 3 Zen 2 Zen 2 Zen 3
1P Flagship R9
TR Pro 3995WX EPYC
MSRP $ 799 $ 3990 $ 5490 $ 5010
TDP 105 W. 280 W. 280 W. 225 W.
Base frequency 3400 MHz 2900 MHz 2700 MHz 2000 MHz
Turbo freq 4900 MHz 4300 MHz 4200 MHz 3675 MHz
Socket AM4 sTRX40 sTRX4: WRX80 SP3
L3 cache 64 MB 256 MB 256 MB 256 MB
DRAM 2 x DDR4-3200 4 x DDR4-3200 8 x DDR4-3200 8 x DDR4-3200
DRAM Capacity 128 GB 256 GB 2 TB, ECC 4 TB, ECC
PCIe 4.0 x 20+
4.0 x56 + chipset 4.0 x120 + chipset 4.0 x 128
Pro features no no Yes Yes

One of the biggest features of Threadripper and Threadripper Pro has been any market that typically uses high speed workstations and can scale the workload. When we talked to a local OEM, the demand for Threadripper and Threadripper Pro from the visual effects industry has been off the chart, where these companies are scratching out their old infrastructure and replacing it again with AMD. This has also been rejected by the recent pandemic, where these studios want to keep the expensive hardware in place and let the artists work from home via remote access.

Threadripper Pro Processors: Four models, three in retail

When TR Pro was launched in 2020, the processors were an exclusive Lenovo for the P620 workstation. The agreement between Lenovo and AMD was not revealed, but it appears that the exclusivity agreement ran for six months, from September to February, and the processors were made available on March 2.

During that time, we tried out one of these workstations for review, and it’s still one of the best modular systems I’ve ever tested:

Lenovo ThinkStation P620 Review: A vehicle for Threadripper Pro

AMD’s first Threadripper Pro platform has four processors, ranging from 12 cores to 64 cores, which mimic their equivalents in Threadripper 3000 and EPYC 77×2, but at 280W.

AMD Ryzen Threadripper Pro
AnandTech Kernels Base
Chiplets L3
TDP Price
3995WX 64/128 2700 4200 8 + 1 256 MB 280 W. $ 5490
3975WX 32/64 3500 4200 4 + 1 128 MB 280 W. $ 2750
3955WX 16/32 3900 4300 2 + 1 64 MB 280 W. $ 1150
3945WX 12/24 4000 4300 2 + 1 64 MB 280 W. OEM

At the top is the 64-core Threadripper Pro 3995WX, with a base frequency of 2.7 GHz and a turbo frequency of 4.2 GHz. This processor is the only one in the family that has a full 256 MB L3 cache, since it has all eight chips completely active. The $ 5490 price is a full 37.5% increase over the Threadripper 3990X to $ 3990.

AMD 64-Core Zen 2 Comparison
AnandTech Thread stopper
Thread stopper
Pro 3995WX
MSRP $ 3990 $ 5490 $ 4425
TDP 280 W. 280 W. 200 W.
Base frequency 2900 MHz 2700 MHz 2000 MHz
Turbo freq 4300 MHz 4200 MHz 3350 MHz
L3 cache 256 MB 256 MB 256 MB
DRAM 4 x DDR4-3200 8 x DDR4-3200 8 x DDR4-3200
DRAM Capacity 256 GB 2 TB, ECC 4 TB, ECC
PCIe 4.0 x56 + chipset 4.0 x120 + chipset 4.0 x 128
Pro features no Yes Yes

At the center of the line is the 32-core Threadripper Pro 3975WX, with a base frequency of 3.5 GHz and a turbo frequency of 4.2 GHz. AMD decided to make this processor use four chips with all eight cores on each chip, resulting in a total of 128 MB L3 cache. At $ 2750, it’s also 37.5% more expensive than the equivalent 32-core Threadripper 3970X.

AMD 32-Core Zen 2 Comparison
AnandTech Thread stopper
Thread stopper
Pro 3975WX
MSRP $ 3990 $ 2750 $ 2300
TDP 280 W. 280 W. 180 W.
Base frequency 3700 MHz 3500 MHz 2500 MHz
Turbo freq 4500 MHz 4200 MHz 3350 MHz
L3 cache 128 MB 128 MB 128 MB
DRAM 4 x DDR4-3200 8 x DDR4-3200 8 x DDR4-3200
DRAM Capacity 256 GB 2 TB, ECC 4 TB, ECC
PCIe 4.0 x56 + chipset 4.0 x120 + chipset 4.0 x 128
Pro features no Yes Yes

The following two processors have no thread equivalents, but also represent a slightly different scenario that we will explore in this review. Both the 3955WX and 3945WX, despite being part of the large Threadripper Pro family, use only two chips in the design: 8 cores per chip for the 3955WX and 6 cores per chip for the 3945WX. This means that these processors only have 64 MB of L3 cache, which makes them somewhat identical to the Ryzen 9 3950X and Ryzen 9 3900X, except that the IO death means that there are eight channels of memory and 128 PCIe lanes here.

AMD 16-Core Zen 2/3 Comparison
AnandTech Ryzen 9
Thread stopper
Pro 3955WX
Ryzen 9
MSRP $ 749 $ 1150 $ 799
TDP 105 W. 280 W. 105 W.
Base frequency 3500 MHz 3900 MHz 3400 MHz
Turbo freq 4700 MHz 4300 MHz 4900 MHz
L3 cache 64 MB 64 MB 64 MB
DRAM 2 x DDR4-3200 8 x DDR4-3200 2 x DDR4-3200
DRAM Capacity 128 GB 2 TB, ECC 128 GB
PCIe 4.0 x 20
+ chipset
4.0 x 120
+ chipset
4.0 x 20
+ chipset
Pro features no Yes no
Cost of motherboard +++

The 3955WX has a higher base rate, but the 3950X has a higher turbo frequency. 3950X is also cheaper and motherboards are cheaper! It may be worthwhile to divide these into a separate comparison review.

The latest Threadripper Pro processor, 3945WX, does not have a price because AMD does not make it available in retail. This section is for selected OEM customers only. Maybe the limited substrate resources in the market right now make it unattractive to make too many of these? Hard to say.

Motherboard: Beware!

Despite being based on the same LGA4094 connector as both Threadripper and EPYC, Threadripper Pro has its own unique WRX80 platform that needs to be used instead. Only selected vendors appear to have access / licenses to create WRX80 motherboards, and your main options are:

All three cards use a transposed LGA4094 connector, eight DDR4 memory slots and 6-7 PCIe 4.0 slots.

Be careful though! It is an option to find an old / refurbished Lenovo P620 motherboard. It is worth noting that Lenovo performs an AMD function for OEMs: processors used in that Lenovo motherboard will be locked for Lenovo forever. This is part of AMD’s guaranteed supply chain process, allowing OEMs to hard-lock processors in specific vendors for the end-to-end security of the supply chain demanded by specific customers. In that case, if you ever want to break down the system to upgrade and sell parts, it is not recommended that you find a Lenovo TR Pro system unless you are buying / selling it as a whole.

This review

The main goal of this review is to test all Threadripper Pro 3000 hardware and compare with the corresponding Threadripper 3000 to get a sense of how much performance is achieved with increased memory bandwidth, or lost due to the small core frequency differences. We also include Intel’s best HEDT / workstation processor for comparison, the W-3175X, as well as the best processors on the market. All systems have been tested according to JEDEC specifications.

Test setup
TR Pro
8×16 GB
DDR4-3200 ECC
TR 3990X
TR 3970X
TR 3960X
4×32 GB
X570 I Aorus
4×32 GB
i9-11900K ASUS
XIII heroes
Copper *
4×32 GB
Xeon W-3175X ASUS ROG
BIOS 0601 Asetek
GPU Sapphire RX 460 2 GB (CPU tester)
PSU Miscellaneous (incl. Corsair AX860i)
SSD Crucial MX500 2TB
* Silverstone SST-FHP141-VF 173 CFM fans were also used. Nice and loud.

Many thanks to Kingston for delivering a complete set of KSM32RD8 / 16MEI – 16×16 GB DDR4-3200 ECC RDIMMs for enterprise testing in systems such as Threadripper Pro.

As part of this review, we also show 64 core processors in 128T mode and 64T mode. This is done to show how some processors can get better performance by having better memory bandwidth per thread – one of the problems with these high core processors is the limited amount of memory bandwidth each thread can access. Also some operating systems (like Windows) struggle over 64 threads due to the use of thread groups.

Source link