The IT world was surprised this week when it became known that almost every processor sold for all types of computers in the past 20 years could be exploited due to two major hardware defects (read our article "What You Need to Know "). . The manufacturers discovered by Google's Project Zero team last year have been researching and working on a solution for months, although the public has only just learned about it.
Due to the nature of Meltdown and Specter, the patches must be at the operating system level and there is a possibility of performance degradation. On the other hand, desktop computers and consumer games may not be as affected as other intensive tasks that are more common in server and database applications.
To update: After our first tests (this article) we investigate a patched desktop system for both Meltdown (patch at operating system level) and for Specter (firmware / BIOS update).
Based on the information received so far, we know that most Intel CPUs are affected. However, this problem also affects selected ARM architectures, while AMD appears to be largely clear. There are three types of exploit, and AMD is vulnerable to the Bounds Check Bypass method. However, this can be remedied by an operating system update and should be associated with insignificant performance costs. The other two variants are reported to have no effect on AMD processors due to different architecture.
Linux is one of the operating systems that received an update to address the exploit prior to release, and the performance of the Phoronix tech site has evaluated performance before and after. They found some pretty debilitating performance degradation in server tests with a Core i7-8700K with a Samsung 950 PRO NVMe storage device. They followed with a Linux gaming test that examined half a dozen titles and largely found that the difference was within the margin of error.
Since Microsoft released an emergency fix for Windows 10 on the same day, we had the opportunity to conduct our own tests. Windows 7 and 8 users should soon be able to get patches directly from the Microsoft website, or wait until Patch Tuesday to get them automatically through Windows Update.
The benchmarks you'll see have all been run in the past 12 hours to determine what impact, if any, this update will have on desktop users' performance. A couple of things to keep in mind: we only had time to test the Core i7-8700K, so no older Intel CPUs have been tested, and we haven't checked whether this patch affects AMD CPUs.
These are all new benchmarks that were run with the same test system under the same conditions. We updated all of the pre-patch data first, then applied and re-tested the patch. Game performance figures are all based on a three-run average. We did things a little differently in the storage tests, as these numbers tend to fluctuate more. We would normally take the average of three runs, but for this test we took the best result for each test and instead showed that. We gave the system 4 attempts to get the best possible result, and between each run that was Shut down the system completely and then restart it. Off to the benchmarks …
Starting with the memory tests, we have the AS SSD benchmark and for these tests I use the Samsung SSD 950 Pro with the 8700K. Here we see very similar sequential read and write results. The numbers after the update are actually a bit better, although we're only talking about a difference of 1-2%.
If we look at the 4K reading result, we see a significant difference. Here we see a very large performance reduction of 23% from a throughput of 44 MB / s to 34 MB / s. However, the random 4K write performance remains unchanged, so that only the random 4K reads are significantly affected.
Interestingly, the write and write performance of the 4K-64 thread was improved with the patch, but the write performance here was increased by 17%. So it's certainly not all bad news, although I would argue that the 4K reading is more serious. The read access time was also 14% lower before the update, but the write access time is largely the same.
In the further course, CrystalDiskMark confirms what was found during testing with the AS SSD benchmark. Here the 4K reading performance after the patch was reduced by 23%. However, the rest of the margins are 5% or less, so nothing is really worth mentioning.
The last memory-oriented test I ran was ATTO Disk Benchmark. Here we find something interesting. These are all sequential tests, so the 4K results here don't necessarily reflect what we've seen before, and clearly they don't. However, if the file size increases to 16 kilobytes, we will notice a noticeable drop in performance with the update. The drop is not as significant as the previously observed 4K reading results, but we see a throughput reduction of up to 9%.
Creation and rendering of 3D content
The Cinebench R15 score is based on an average of three runs. We see a slight fluctuation in performance here, but nothing to worry about. The multi-thread score is reduced by 2%, while the single-thread score is increased by one percent.
We look similar in Blender, the update came ahead in a fraction, but again within the error limit.
Another render test, this time the Corona benchmark and nothing to report here. So it seems that your rendering workloads are not affected.
Likewise, your spreadsheets remain as snappy as ever, we see no effects here at all.
The 7-Zip compression and decompression performance looks similar. This is also within the error limit for this test.
VeraCrypt also saw no difference in the 1 GB and 50 MB AES encryption and decryption tests.
Okay, time to get serious with a few gaming benchmarks. First of all we have the always reliable ashes of singularity and here we see a small increase in performance after the patch has been applied. Admittedly, the 1% low result was only improved by 1.7%, but it is certainly not slower. We are also tied to the CPU using the high quality preset, not the GPU.
If you increase the quality preset to crazy results in a GPU-limited scenario, you will find similar results before and after installing the patch.
Assassin's Creed Origins was a game I really wanted to try because it offers protection against piracy. As you can see, there is nothing to talk about here, apart from the fact that the patch has no negative effects on frame rates and switching to the ultra-high quality preset doesn't change anything.
Then we have the Battlefield 1 results and using the medium quality settings at 1080p with the GTX 1080 Ti will achieve similar performance before and after the patch. After the patch, we even see a very slight improvement. This is also evident in the settings for the ultra quality, so that everything looks good with the Core i7-8700K.
Conclusion so far
Well there you have it. Desktop users don't have to worry much about performance degradation, especially gamers. We haven't tested older CPUs yet, but given the type of workloads affected by the patch, I don't think there will be problems with desktop hardware, but we'll definitely report back if there are any.
The decrease in 4K read performance for high-speed NVMe drives is worrying. While this shouldn't affect games, applications that respond to them may experience performance degradation. Of course, the short list of applications I tested did not really shorten the performance period.
However, the problem persists and has a far greater potential for impact on servers. It is a serious problem for data centers, both in terms of performance and, above all, security. This is not our area of expertise or interest, so we leave these tests to those who are better equipped to do it.
Update (1/5): When we tested and published this article, the Windows 10 emergency fix was unavailable for just a few hours, and no microcode or firmware updates were available. For laptops and brand name desktops, this means that you will also need to update your system with new firmware from your respective OEM. For builders, this means waiting for an update from your motherboard manufacturer.
At the time of writing, a handful of companies have started offering these updates for some of their product lines: Dell, Lenovo / ThinkPads, Asus, Microsoft Surface, and some others. When our test platform receives a security firmware update, we plan to add these results to this article. Internet users have started to publish benchmarks of their systems with both patches, and the result seems to be a bit worse performance than if they only applied the operating system patch.
As we understand it, the Windows patch and other operating system updates only cover one of three known vectors to take advantage of the bugs. The combination of operating system and firmware should cover all three areas. However, in some scenarios, application level patches are also required to be completely secure. To better understand Meltdown and Specter, we recommend that you read our brief information or read the white papers.
Update (1/7): After our first tests, we investigate a patched desktop system for both Meltdown (patch at operating system level) and for Specter (firmware / BIOS update).