The IT world was surprised this week when it was revealed that almost every processor sold in the last 20 years that powered any form of computer could be exploited due to two major hardware flaws (read the article). ‘What do you want to know’).
Discovered last year by Google’s Project Zero team, the manufacturer has been investigating and improving for months, though the public is now only aware of it.
Due to the meltdown and the nature of the specter, the patch should fall to the OS level and is likely to lose performance.
On the other hand, for consumers, desktop computers and games may not be affected as some of the more intensive tasks for applications, servers, and databases.
Update: Based on our initial testing (this article), we will take a more in-depth look at modified desktop testing for both Meltdown (OS level patches) and Spector (firmware / BIOS updates)
Based on the information received so far, we know that most Intel CPUs are affected. But the issue also extends to the selected ARM architecture, while AMD appears to be largely unambiguous.
There are three types of exploits, and AMD is vulnerable to the “bounds check bypass” approach, but can be remade through an operating system update and should have a smaller performance cost.
The two other variants are reportedly not affecting AMD processors due to differences in architecture design.
Linux was one of the operating systems that was updated on the exploit before it appeared, and the technology website with a performance benchmark comparable to the Phoenix website before and after.
They got poor performance when running server-related tests using a Core i7-8700K with a Samsung 950 Pro NVMe storage device.
They followed up with Linux gaming test results, which looked at half a dozen titles and most of them found that the difference was within the range of the bug.
With Microsoft rolling out emergency fixes for Windows 10 on the same day, we have our own trial opportunity. Windows 7 and 8 users should soon be able to get their fixes directly from the Microsoft website, or wait until Patch Tuesday to be able to get them automatically. Windows update
The comparison you’re going to see has been done in the last 12 hours to see how this update affects the performance of desktop users.
Something to note: We only had time to test the Core i7-8700K, so no older Intel CPUs have been tested, and we haven’t checked to see if this patch affects AMD CPUs. is.
Here are all new benchmarks that were run on the same test system under similar conditions. We first update all pre-patch information, then apply the patch and test again.
All gaming performance figures are based on three sprint averages. For repository testing, we have done some work because these numbers are more volatile.
Usually we use an average of 3 runs, but for this test we got the best results for each of these tests and showed it instead.
We tried to give the system 4 times to get the best results possible and during each run the system completely shut down and then a backup boot.
Let’s start with storage testing, we have an ASD benchmark and for these tests I used the Samsung SSD 950 Pro with 8700K.
Here we see very similar sequential read and write results. The number after the update is slightly better, although we only mention a 1-2% difference here.
Where we can see a significant difference is that when looking at 4K reading results here, we found a dramatic 23% reduction in performance from 44 MB / s throughput to 34 MB / s.
The random 4K writing performance is unchanged, however, this is just a significant drop in random 4K readings.
Interesting, though 4K-64 threaded read and write performance is improved with patches. But there is a 17% increase in writing performance here, so this is not all bad news, although I would argue that the consequences of reading 4K are more severe.
Access time before update is still 14% shorter. But writing time is too much.
We got a CrystalDiskmar. 4K read performance dropped 23%, confirming what was observed when tested with the ASD SS benchmark, the remaining margin was 5% or less, though nothing notable.
The last storage-focused test I used was the ATTO Disk Benchmark, and here we found something interesting.
These are all sequential tests, so the 4K results here do not necessarily reflect what we had seen before and were not clearly shown.
However, as the file size increases to 16KB, we begin to notice a noticeable drop in performance updates. This decline is not as significant as the 4K reading result seen earlier. But we found a 9% drop in throughput.
