ASRock Rack X570D4I-2T Benchmarks
We wanted to give some sense of relative performance to more traditional server solutions on the market.
Python Linux 4.4.2 Kernel Compile Benchmark
This is one of the most requested benchmarks for STH over the past few years. The task was simple, we have a standard configuration file, the Linux 4.4.2 kernel from kernel.org, and make the standard auto-generated configuration utilizing every thread in the system. We are expressing results in terms of compiles per hour to make the results easier to read:
There are many folks out there who offer that the Xeon E3-12xx series is still very fast. Likewise, older Xeon E5 V1-4 systems are extremely competitive. We wanted to show a fairly wide range of CPU options cutting across families here. As you can see, the Ryzen 9 3950X performance is significantly better than older generation systems even in this mITX form factor.
c-ray 1.1 Performance
We have been using c-ray for our performance testing for years now. It is a ray tracing benchmark that is extremely popular to show differences in processors under multi-threaded workloads. We are going to use our 8K results which work well at this end of the performance spectrum.
Here we can see c-ray 8K results that are solid. Performance of AMD Zen, Zen+, and Zen 2 chips tend to be great on this type of benchmark. If you wanted to build out a render farm and still have manageable nodes, then this may make a lot of sense as a solution.
7-zip Compression Performance
7-zip is a widely used compression/ decompression program that works cross-platform. We started using the program during our early days with Windows testing. It is now part of Linux-Bench.
7zip shows something intriguing. Here, the performance of the Ryzen 9 3950X is well above that of the much more costly Xeon D-2183IT. The Xeon has RDIMM support so it can hit more appropriate memory capacities for a 16 core CPU, but in terms of raw CPU performance, the Ryzen chips are much faster than Xeon D 16-core offerings.
OpenSSL is widely used to secure communications between servers. This is an important protocol in many server stacks. We first look at our sign tests:
Here are the verify results:
Here we just wanted to highlight how much faster the Ryzen 9 3950X is versus the Xeon E-2200 line. Simply having more cores and cache helps a lot. Performance per core on the Xeon E-2288G is very close, but we have twice the cores with the Ryzen 9 3950X.
Another key one here is that the Ryzen 9 3950X is about twice as fast as the Xeon E5-2670 V1’s. The Xeon E5 series has more PCIe and memory capacity, but if one is space/ compute-bound, the Ryzen 9 3950X in the ASRock Rack X570D4I-2T is similar to four E5-2670 V1’s.
Chess is an interesting use case since it has almost unlimited complexity. Over the years, we have received a number of requests to bring back chess benchmarking. We have been profiling systems and now use the results in our mainstream reviews:
Here we focused on more server CPUs. This is where we wanted to highlight something that is extraordinarily important. Other than the other Ryzen CPUs and Xeon E-2288G on this list, the remainder of this list has ECC RDIMM support. Platforms such as the EPYC 7282 also have significantly more PCIe Gen4 connectivity and memory bandwidth.
Memory capacity is a big deal. Generally, with 16-core processors, we would expect to see a minimum of 128GB-256GB (8GB-16GB per core) paired with them for general virtualization workloads. Most of the single socket EPYC nodes we deploy are 256GB of memory at 16 cores and go up from there. That is a weakness of using mITX and a consumer platform even if performance is great.
Overall, performance is good if all you are looking to obtain is a lower core count and memory footprint server.
We may have readers who want to know the performance of the chip on the X570D4I-2T versus traditional workstation motherboards. The numbers are very close. As a control, we tested this motherboard against an ASUS PRIME X570-P that we use for storage testing. Looking across two dozen benchmarks, the variance was +/- 1.5% which is within a margin of error. We felt this was close enough that it is less of a factor than the raw price/ performance this combination offers.
While the X570D4I-2T and its predecessors the X470D4U series supports ECC memory, there is a bit of a gotcha. As readers noted in the original X470D4U reviews, while ECC memory is supported and performing error correction, the reporting of that error correction was not functioning. In other words, even if you were experiencing continuous memory errors, no log of those errors was being recorded in the IPMI event log where one might expect them to show up. A user over on the Level One Techs forums had a conversation thread with someone from ASRock Rack, who reported that while the AM4 platform had ECC support, it did not have error reporting support.
For the X570D4I-2T, we verified that operating systems (Windows and Linux) were reporting that multi-bit ECC support was enabled, but we were unable to test if the error reporting situation had changed. My personal assumption is that it is likely still not working, as the problem seemed to stem from the CPU memory controller not providing the ECC error reports; the chipset likely has very little to do with the process. Still, on the current BIOS revisions, the X570D4I-2T runs a newer version of AMD’s AGESA code for the chip (22.214.171.124) compared to the X470D4U (126.96.36.199abba), so it is possible things have changed, but it simply is not something we are able to verify.
We wanted to push the X570D4I-2T as much as we could, and so we picked the biggest and most power hungry Ryzen CPU we could.
- Idle Power: 29W
- STH 100% Load: 174W
- Maximum observed power: 183W
These results were observed on 120V power using a basic Kill-A-Watt meter. The system is powered by a consumer-grade 80Plus Bronze power supply similar to what is found in many low-power short-depth server chassis.
With essentially nothing else plugged into the system except the CPU, cooling fans, and a single SSD, these numbers simultaneously represent the worst-case power draw thanks to the CPU, but also the best case idle power because almost nothing else is attached to the system.
Next, we are going to discuss some of our closing thoughts around the solution.