Search the Community

Adblock: Google did not slow down and lag YouTube performance with ad blocker on by Sayan Sen Back in November, the internet was abuzz with conspiracy theories of Google purposely slowing down YouTube on Mozilla Firefox while its own browser, Chrome, would work fine. However, those rumors were shot down soon after as Google explained that it was not just Firefox that was the subject of "suboptimal viewing" experience and that it was happening irrespective of the browser. Google put the blame on "installed ad blockers." Fast forward two months, now in January we just had another similar incident with several user reports online of YouTube being slow with adblockers. As is usually the case in these instances, most netizens of course were quick to once again point the finger at Google. The issue was first brought to attention by Reddit users and the thread blew up with many more chiming in to say they were experiencing a similar thing. As it turns out, the bug was not YouTube or Google and was in fact a problem with Adblock and Adblock Plus' recent update Version 5.17.0. Hence if you were using something else like uBlock Origin (like me), you probably did not notice any slowdown. Adblock Plus developers noted the issue as performance regression on its GitLab repo. It wrote: Recently, ABP released version 3.22 which upgraded the bundled extension engine version to 1.1.1. AdBlock released version 5.17.0 which also updated the extension engine to version 1.1.1. We've had several reports of slow response time since the update. It appears to be an issue in the extension engine since ABP, AdBlock, and the EWE test extension all seem to experience a similar issue with the 1.1.1 version of the extension engine. The issue has now been fixed with the latest eyeo's Web Extension Ad Blocking Toolkit (EWE) version 1.1.2, where the breaking change has been reverted. Name: @eyeo/webext-ad-filtering-solution New version: 1.1.2 Reverted "Content filters are now updated via the history.pushState() event, when single page apps navigate using the browser's history API (EE-14, EE-90)". Breaking changes: None. Other changes: Reverted performance regression. Hence, if you are one of those users who is experiencing issues with Adblock and Adblock Plus, you should update your extension. It is also advised not to run multiple content filters or adblockers on your browser if you are facing performance issues. As an online publication, Neowin too relies on ads for operating costs and, if you use an ad blocker, we'd appreciate being whitelisted. In addition, we have an ad-free subscription for $28 a year, which is another way to show support!

Following Microsoft Defender, Avira now freezing Windows at boot with 100% CPU / RAM usage by Sayan Sen Just a few days ago we published AV-TEST's and AV-Comparatives' findings regarding system performance impact by various anti-malware products. In the article, we covered how Microsoft Defender, AVG, McAfee, and Norton, among others, can hog Windows PCs under different circumstances. This was just a couple of days after we covered an issue with Microsoft Defender causing performance loss on Windows 11 23H2, where the workaround was to reset Defender. However, if you think third-party anti-virus solutions work flawlessly, you will be quite incorrect. According to multiple reports online, a recent Avira update is leading to the entire system freezing right at startup when the anti-virus app kicks off, and this is seemingly happening on Windows 10 as well as Windows 11. Most of the complaints online testify that a similar thing is happening. While the system is responsive immediately when it boots up and gets into Windows, it freezes up and becomes unusable and unresponsive as soon as the Avira application loads up. This is happening on SSDs too, whether NVMe or SATA, and the situation is likely worse on hard disks (HDDs). Some users were able to launch Task Manager in the small window of time they got when the system booted, which led them to notice that the CPU and RAM usage were seemingly both pegged at 100% as the system came to a total standstill. The issue has been pretty much confirmed to be something related to Avira as more and more users were able to fix the problem by uninstalling Avira. If you are one of those affected too, you can do so by rebooting Windows into Safe mode and then uninstalling the Avira antivirus program from the Control Panel app > Programs and Features > Uninstall a program. In case you are worried about malware protection, Microsoft Defender should take over from there when Windows notices you don't have an anti-virus enabled. The Avira Support X (Twiiter) handle has yet to publish any update and the latest Avira module release note also does not say anything about the issue yet: Avira Security 1.1.96 Bugfixes Stability Improvements Avira Browser Safety 4.2.7 Bugfixes Avira Phantom VPN 2.43.1 Bugfixes Avira Password Manager 2.19.13 Bugfixes Avira Speedup for Windows 6.26.0.18 Maintenance Stability Improvements Avira Secure Browser x86(32-bit) 118.0.22916.118 Avira Secure Browser x64(64-bit) 118.0.22914.118 Bugfixes Stability Improvements The issue happened over the weekend which is likely why Avira has not been able to get to the root of it yet. And in case you are thinking of switching over to something else, you certainly could, though bear in mind that in the latest malware protection ranking, It was amongst the best performers.

Windows 11 23H2 upgrade causing performance loss and trusty Microsoft Defender may be why by Sayan Sen This weekend we published a feature looking at the performance improvements, or regressions, one can expect if they were to move from a clean-installed Windows 10 22H2 system to a clean-installed Windows 11 23H2 PC (similar to the one where we upgraded to Windows 11 22H2). What we generally noticed was that the differentials were typically within the margin of error or in the imperceptible zone, though there were situations where one was quite a beat ahead of the other. Although we haven't yet published our findings showing the differences between Windows 11 22H2 and 23H2 (we plan to do it soon), several users on the internet say they are suffering from performance losses after upgrading to version 23H2. Reddit user BNSoul opened a thread on the Windows 11 subreddit explaining how they noticed performance regression of around 5 to 8% in several CPU benchmarks. Games also apparently exhibited random stuttering and hitching. They write: To put it simple[sic], every CPU benchmark shows significantly reduced CPU performance after updating to Windows 11 23H2 from 22H2, even after a fresh/clean install. ... I could add an endless list of benchmark results here but just lets say it's always 23H2 5-8% slower in every single one be it single or multi-thread compared to 22H2. Games are also affected with random stuttering, all of this fixed by rolling back to 22H2. In a separate thread on the Microsoft forum, user Anant Acharya posted about similar performance degradation issues from the CPU side wherein games would randomly drop frames due to the GPU utilization dropping off in those instances. They also report texture loading issues. Curiously, the CPU usage seems to remain the same but the game issues bandwidth-related error messages. After I had updated to the Windows 23H2 update. I have been noticing sudden stutters and drastic FPS drops in the above-mentioned games. .. In newer games like Forza Horizon 5 the GPU utilization (remaining at 92-99 %) suddenly drops to 20-23 percent and the FPS drops from 70-80 to about 10-23, textures not loading. while the CPU utilization remains the same. the game returns a Bandwidth issue warning. as the game was installed on my 7200 RPM hard drive. i decided to move it to my SSD. the issue still existed. .. Call of Duty Modern Warfare 2 (campaign remastered) crashed with GPU errors after updating the latest drivers and re-installing windows entirely (still 23H2) the issue presists[sic]. as it has suddenly been brought up and i see no physical signs of damage that may have caused it. and the issue being brought up right after the 23H2 update. i suspect the OS to be the cause. Several others upvoted this query and also posted their own findings and issues there. Fettman 53 writes: Unfortunately I have nothing to add, other than that I'm having the same issues. Games I played literally earlier the same day before downloading 23H2 worked flawlessly, and after downloading the OS update, they now run like garbage and have the same issues you listed here. A friend of mine is reporting the same. David Alfredo, another user who ran several benchmarks, posted: Same issue here after a clean install of Windows 11 Pro 23H2, I spent a whole day troubleshooting and noticed a CPU performance degradation in all the usual benchmarks (Cinebench, Geekbench, 7-zip built-in benchmark, Corona renderer, OCCT, Super Pi, Y-Cruncher.....) so yeah, it seems like there's something going on with the latest update that affects CPU performance to the point that games now show stuttering. Maybe it's just me but there's not a single CPU benchmark that gets the same or better results on 23H2 compared to 22H2, they all score much lower. Although the two threads, the Reddit one and the Microsoft forum one, may not stem from the same root cause, there seems to be a solution for the former, which involves resetting the Windows Security app, which seems to resolve the performance problems users may be noticing after upgrading to Windows 11 23H2: Received an answer from Microsoft after 3 days, they told me to reset Windows Defender through a couple of PowerShell commands (1- "Set-ExecutionPolicy Unrestricted" and 2- "Get-AppxPackage Microsoft.SecHealthUI -AllUsers | Reset-AppxPackage") then reboot and enable CPU Virtualization in BIOS (SVM in my X570 BIOS), then in Windows 23H2 open Windows Security and enable Memory Integrity under the Core Isolation settings. Restart and Hypervisor should be running, Virtualization Security will be Enabled and... that fixes the CPU performance issues, CPU now performing as in 22H2 where I had these security features disabled. Tested some benchmarks and games, everything is now ok within margin of error compared to 22H2, GPU benchmarks are 3-5% faster which is nice, games are marginally faster at least CP2077 and SoTR benchmarks, CPU benchmarks on the other hand some performs the same as in 22H2, others improved and a couple of them are maybe 0.2% (margin of error) slower perhaps due to memory integrity being enabled. Hence, if you are experiencing performance-related issues on your newly upgraded 23H2-based PC, perhaps it's worth a shot to reset Defender.

Microsoft's default BitLocker on your Windows 11 PC is hitting even the fastest SSDs hard by Sayan Sen Earlier this month, we covered an issue about a "65000" BitLocker encryption error message which was affecting both Windows 11 as well as Windows 10 PCs. And while that was only a reporting issue, something else is actually impacting the performance of Windows 11 Pro PCs. Tom's Hardware has found that the default software-based BitLocker encryption (XTS-AES 128, where XTX stands for XEX Tweakable Block Cipher with Ciphertext Stealing and AES is Advanced Encryption Standard), which is enabled by default on pre-built PCs like laptops and notebooks, among others is hitting the performance of SSDs pretty hard. The outlet notes that the impact can be as bad as up to 45%. This report comes hot on the heels of another one where we saw that Linux (Ubuntu) has continued to grow its lead over Windows 11 and in some cases, it is more than twice as fast. Tom's Hardware says: We reached out to several OEMs, and Dell, HP, and Lenovo told us they ship systems with Windows 11 Pro with software-based encryption unless a user orders an SSD that has hardware-based encryption available. It's not clear if they always enable hardware encryption on every SSD that supports the feature, but if you don't pay extra for such a drive, you'll likely end up with reduced storage performance. We haven't received a response from several other OEMs yet, but we suspect most have similar policies in place. To test the performance hit, one of the fastest PCIe Gen4 NVMe SSDs in the form of the Samsung 990 Pro 4TB was used. The least impact was seen in the sequential peak speed tests. In a DiskBench 50GB file copy test, the drive with software encryption had a 13% performance loss since unlike the hardware-accelerated Opal solution, the software one uses the CPU to do the encryption. The performance hit is far worse on random reads and writes though especially in the Queue Depth 1 (QD1) tests. In the 4KB QD1 random read test, there is a 21% slowdown in the IOPS (input/output operations per second). The impact is even greater in the case of random writes as the 990 Pro sees a 46% hit compared to hardware encryption and a 43% hit compared to no encryption: When the QD is changed to 256, the impact is different as hardware and software both see big slowdowns in performance when talking about writes: The 4TB 990 Pro tested here comes with 4GB of LPDDR4 DRAM cache as well and despite that, the slowdown in randoms is very prominent. You can find the full test results at the source link below. If you have an OEM pre-built system that came pre-installed with Windows 11 Pro, chances are it has software BitLocker enabled. You can check that by running an elevated command prompt (CMD as an admin) and entering the command: "manage-bde.exe -status". When you do so, the Protection status will be displayed as "Protection On" and in the case of software encryption, the encryption method is set to "XTS-AES 128". A device with BitLocker disabled will display "Protection Off" and the Encryption method will be "None". Another way to do so is to launch the Disk Management console and check to see if the drives show "BitLocker Encrypted" in brackets. Source and images: Tom's Hardware

Old Intel 10th Gen is snappier vs newer 13th Gen and also AMD's Ryzen 7000 in Windows 10 by Sayan Sen Since July, we have been covering a series of performance tests that examine the responsiveness and snappiness of Intel's newer processors on Windows 10 and Windows 11 compared to some of the older parts. Although it is no surprise that Intel's 13th Gen/12th Gen SKUs are easily able to outdo their 10th Gen counterparts in benchmarks, the idea behind these tests has been to quantify the responsiveness or snappiness of the processors in day-to-day PC activities. One of the key components of a snappy experience is the I/O and apparently, certain architectural decisions made at Intel with 12th Gen and 13th Gen processors, which are Intel's first hybrid parts, have apparently led to the company moving the "I/O driver" of the uncore (System Agent) to be moved off the CPU. (On a related note, anti-virus apps can also slow down certain aspects of your PC as noted by AV-TEST in its latest report.) This has led to performance penalties in the case of Windows 10 it seems, though the specific Thread Director optimizations on Windows 11 meant that the latter is typically more responsive of the two OSes on a 12th Gen (Alder Lake) or 13th Gen (Raptor Lake) system. To confirm if indeed the newer Microsoft OS is not as reactive as Windows 10, another study, this time with the 10th gen Core i9, was run and the results showed Windows 11 was definitely lagging behind, at least by a slight bit, in almost every single test. While most of these tests were done using Intel 10th Gen and 13th gen CPUs, an AMD Ryzen 7000 series processor, based on Zen 4, was thrown into the mix this time, and curiously enough, AMD's much more modern offering also fell behind the older Intel chip more often than not. The CPU used for testing was the Ryzen 9 7950X, which is a 16 core 32 thread SKU. The Ryzen managed a victory in one out of the four tests as it was the best chip at opening music files. The Zen 4 recorded took an average of 72.2 milliseconds compared to 112.5ms on the 13900K and 124.9ms on the 10850K. Interestingly, the 10th Gen Intel CPU looks like the most consistent of the bunch as its results were grouped the closest together. In terms of visible input latency with a mouse as the input, the numbers were the closest between the three indicating none of the CPUs is a bad choice for such a scenario. DPC (Deferred procedure call) latency though is the best on the 10th Gen Intel, but the Ryzen 7950X is still easily able to outdo the 13th Gen processor. The worst showing from AMD was in file searching using the Taskbar Search. In this case, both the Intel CPUs were nearly five times faster as the 7950X took over 80 ms to complete the task as compared to the Intel chips that were able to do it under 17ms. Head over to this article to compare and contrast with the Windows 11 data. You can watch the full video on Tech YES City's channel on YouTube at the source link below. Source and images: Tech YES City (YouTube)

Windows 11 vs 10: Intel CPUs apparently sluggish, and it's worse without Microsoft's help by Sayan Sen Update: More related testing suggests Microsoft's Windows 11 is actually generally slower on Intel non-hybrid parts. Intel moved to a hybrid x86 CPU design for its 12th generation of Core processors (Alder Lake). Not only that, but the company also made several other changes to other components too, dubbed the "Uncore" portion of the processor, also referred to as the System Agent, which includes the Northbridge. One such change is apparently causing various latency issues and sluggish performance in general on Intel's hybrid chips according to a report by YouTuber Tech YES City (TYC). TYC, citing industry sources with knowledge of the matter, explains that the problem, apparently, is a consequence of the "I/O driver" of the uncore (System Agent) that was allegedly moved off the CPU on Intel's 12th Gen and 13th Gen chips. Using latency tests, TYC shows that the problem is especially pronounced on Windows 10 in most cases. Although Windows 11 is not fully immune to these performance quirks, Microsoft and Intel's combined efforts on optimizing Thread Director for the hybrid processors do pay off in some of these cases, even though past gaming performance numbers have not suggested the same. In the description of their YouTube video, TYC writes: I always thought Intel's CPUs were snappy, in fact that were the snappiest of snappiest, even vs Ryzen, where AMD have improved a lot over the years with their latency. However with the advent of 12th gen Alder lake and 13th Gen i9-12900k and 13900K CPUs respectively, Intel took something AWAY, and that was they moved the I/O driver directly off the CPU (that's what I was told from multiple sources), and this was the exact reason why latency issues have been occurring on these CPUs for windows users In order to test and understand the validity of the perceptive difference they felt, TYC compared their 10th Gen Core i9-10850K (Comet Lake-S) with the Raptor Lake-S i9-13900K (13th Gen). In some cases like DPC (Deferred procedure call) latency, the performance is bad on both Windows 11 and 10, though still, the Windows 10 system does do worse. This makes sense as DPC is directly related to driver handling efficiency and some part of the I/O is involved. Other cases like dragging and dropping MP4 files and rapidly opening MP3 files in Adobe Premiere Pro also exhibit poor showing on the part of the newer Raptor Lake CPU. Some of the performance differences seen here may also have to do with the core-to-core (c2c) latency differences in Raptor Lake compared to Comet Lake. As you can see in the images above, courtesy of AnandTech (1,2), the c2c latency varies quite a lot on the hybrid Intel 13900K compared to the 10900K. The c2c on the 10900K varies from as low as 5.6 ns to as high as 23.8, while that on the 13900K is anywhere from 4.0 ns to 53.9, and the latencies between the E-cores are especially bad. Source and images: Tech YES City (YouTube)

Latest Windows 11 22H2 Patch Tuesday (Moment 2) may be slowing down SSDs, boot times by Sayan Sen This past week, Microsoft released Patch Tuesday updates as it was the second Tuesday of the month. It was also with this cumulative update that Moment 2 got pushed to all users; and according to some users' feedback, the update may be slowing down SSDs or other boot drives. On a Reddit thread, a user "mesp21" says their NVMe SSD speeds were more than halved, going from the rated 7,000MB/s down to as low as 3,000MB/s. The issue seemed to get fixed when they reverted back. They write: Tanked my SSD nvme reading and writing speeds, like A LOT. Went from 7000 to 3000, sometimes 1000 using the balance energy profile in my Legion 5 2021. Just uninstalled the cumulative update and my SSD is reaching the 7000 read speed again and Windows is snappy again. This comment has quite a few upvotes implying that at least a few of the those users also experienced similar issues. Some of the others on the thread have responded to this comment saying how their game load times were "noticeably" slower, and that there was "stuttering" issues post-Patch Tuesday installation. Another user "Mrbigdog99" has also posted a screenshot showing CrystalDiskMark run before and after the update. The sequential writes do seem to suggest a massive slowdown: Another user writes they faced similar problems, like slower boot times, though this time, it was with the initial Moment 2 release (KB5022913). The user sebascq writes: Hello guys! so I updated my laptop to the 22H2 SO 22621.1344 and noticed that the boot up time increased by a lot it went from around 14 seconds to 31 seconds which is weird Microsoft has confirmed that a similar file copy slowdown bug has been fixed with Moment 2, and it is possible that this is a separate problem entirely. Interestingly, the NVMe speed issue isn't something new as Windows 11 has been impacted by such problems in the past too.

Microsoft seemingly enabling VBS by default in Windows 10 too, leading to performance loss by Sayan Sen Time and again, Microsoft has stated that Windows 11 is the best OS for gaming. First in mid-2022, when DirectStorage for PC was first announced, and more recently, earlier this year, with a video that was perhaps a little awkward; and this was just days after testing showed nearly no discernible difference between it and Windows 10, at least in terms of performance. Reports have suggested in the past that certain security features which are enabled by default in Windows 11, like Virtualization-based Security (VBS) or Core Isolation/Memory Integrity, can lead between massive to noticeable performance loss, even on officially supported Windows 11 CPUs (Microsoft has itself recommended disabling it). And now, it looks like Microsoft is also enabling VBS by default on Windows 10 22H2 as well. Tom's Hardware's Avram Piltch noticed that Memory Integrity was enabled on their clean-installed Windows 10 build. The publication also conducted a round of testing across 15 titles at four different game settings and resolution. These gaming tests were run on a Windows 11 system powered by the AMD Ryzen 7950X3D, which has already proved itself to be the most efficient gaming CPU, completely blowing Intel out of the water, and an Nvidia RTX 4090. While one would expect the biggest performance deltas at 1080p, which is the most CPU bound resolution out of all the three, it is really hard to predict, as it seems performance is all over the place depending on the settings used. Overall, it is a 5% drop with VBS on. You can view the full performance chart in the image above. Although this test was only on Windows 11, it is likely that VBS is going to have similar impact on Windows 10 as recent tests have showed performance on the two OSs are in the same ballpark. Microsoft isn't the only company which is having to deal with performance loss at the moment though, as both AMD and Nvidia have also had buggy drivers recently. On the former, a rare bug in the driver is causing over 50% performance loss, meanwhile Nvidia just fixed a high CPU usage bug with its latest Game Ready driver. Source and image: Tom's Hardware

Windows 11 AMD driver bug causing over 50% performance hit but luckily it's rare by Sayan Sen Earlier today, we published a deals piece on AMD's top end cards, the RX 7900 XTX, XT, and the 6950 XT, simply because how much better these are priced compared to their Nvidia counterparts. So while the hardware side is excellent and hard not to recommend, the software side can be a very different story, as quite often it is a one step forward two steps back for the GPU drivers. With the latest driver version 23.3.1, AMD fixed a major annoying issue that would cause the display driver to crash and timeout, leading to the bug report tool popping up, though older cards still remain affected which means such users will have to make do with the MPO workaround. While that is great, the latest driver is also apparently causing major performance loss by 50% or more, at least for a small subset of users. The performance issue was noticed on 3DMark Time Spy, which is a synthetic benchmark based on DirectX 12. Steam user "opinali" reported the performance drop on their Windows 11 22H2 PC running a Radeon 7900 XTX alongside a Ryzen 9 5900X. They write: I have a massive drop in the TimeSpy scores (both regular and extreme) in a Ryzen 5900X + 7900 XTX. Example: https://www.3dmark.com/compare/spy/35755350/spy/36463560 Both tests with same general config, CPU is auto-overclocked by Ryzen Master but GPU is stock. Important detail that I made this test after upgrading the Adrenalin driver to 23.3.1 released yesterday, so this could be a driver problem but I ran many other benchmarks, from 3DMark and others, also played some games, nothing else drops performance, only the TimeSpy which is a cataclysmic >50% loss. In response, a 3DMark developer stated that they were unable to reproduce the issue themselves, though they noticed that around 3% of the Time Spy users did have performance drop right after they updated to the latest 23.3.1 driver. Update on this: We have not been able to reproduce this in our test lab at this time. We're still trying, but nothing so far. So we know at this time that this is not a general issue that affects everyone with 7900XT/XTX. If this were such an issue, it would have also come up during driver approval testing but that process saw nothing odd with the latest drivers when they were tested. But we also looked at our database to compare results on the previous driver vs. new driver on any result using 7900XT or XTX and can confirm that this appears to be a real, if very rare issue. Among all results with the new driver, approximately 3% of the results show abnormal (very low) scores on Time Spy. No similar group of very low scores appear on the results with the previous driver version. This would lead us to believe that this is some kind of issue with the latest driver that requires some additional conditions to be true to appear. .. Since benchmark itself has not been modified for ages, our current view is that this is a rare driver bug that somehow causes the system to run in some kind of low power mode while Time Spy is running, which then tanks the performance to a very low level. UL benchmarks has confirmed that AMD were able to reproduce it and it does look like an issue on AMD's part. However, the issue may be present only on the Windows 11 Release Preview Insider channel which could explain the low number of affected user cases. Note that release preview channel Windows is not officially supported by us. We only test on it with a single system so any hardware-specific issues are not visible to us unless they occur with the "live" version of Windows. In other news, AMD has apparently reproduced this internally and it is starting to look more and more like a driver issue, so at this time I can only offer "rollback to previous or ignore and wait for future driver fixes". While this driver bug is known to affect only 3DMark for now, it is possible that some games may also be affected, although the performance hit in games is unlikely to be as bad since Time Spy is a synthetic workload. Nvidia are also not immune to issues as recently, it was found to increase CPU usage. Thankfully, it has since been fixed. Via: VideoCardz

After all the Windows 11 security touting, Microsoft Defender comes last in AV-TEST's result by Sayan Sen Ever since Windows 11 came out, Microsoft has made a big fuss over why its new operating system is the most secure Windows OS ever. The company constantly nudges users who are on older Windows versions, like Windows 10 21H1 or Windows 8.1, to move over to Windows 11. However, anti-malware assessment firm AV-TEST found that the performance of Microsoft Defender was pretty mediocre in its latest August 2022 report. This was the first such test conducted on Windows 11 and Defender came in last place as it scored just 16 points. Thankfully for Microsoft, it was not alone here, as PC Matic also got the same score. You can view the full breakdown of the scores as well as the product certification they received in the images below: The results are certainly a little surprising since Defender generally did really well on Windows 10. In fact, it often came in first place with the full 18 points. Over on Windows 11 however, it received 5.5 out of 6 in "protection" and 4.5 in "performance". Only in the "usability" category did it manage to get the full 6 out of 6 points. As such, it got the AV-TEST Certified rating whereas previously on Windows 10, it generally received Top Product certification. According to AV-TEST, Defender saw the biggest impact when copying files, both locally and in a network environment. This is why it scored so low in the performance section of the test, and it is not the first time for Microsoft Defender either, as AV-Comparatives too had discovered similar issues before. Source: AV-TEST

It's last place again for Microsoft Defender, this time in AV-Comparatives' Windows 10 test by Sayan Sen Earlier today, AV-TEST, which is a reputed anti-malware assessment firm, released its latest rankings for anti-virus products for home users. This was the first time that AV-TEST evaluated the performance of such products on a test bench running Windows 11. And surprisingly, the overall performance of Microsoft Defender turned out to be far worse than expected. In fact, it only managed to secure the last place where it shared the position with another product. This is quite contrary to how Defender generally does on Windows 10 in AV-TEST's assessments. The findings of AV-TEST seem to match AV-Comparatives' latest test results as well. In the case of the latter, Microsoft Defender was found to bog down a user's system the most (according to a quantitive metric) out of all the tested products. File copying, archiving, and app installations seem to be the Achilles' heel of Defender as they have the greatest performance impact according to AV-Comparatives' test, and once again, it is in line with what AV-TEST also found, and extremely similar results were also seen in AV-Comparatives' own previous ranking. AV-TEST graded the different anti-virus products using its own AV-C or AVC score as well as a PCMark score. The AVC score has the full mark of 90 while the PC Mark score is out of 100. The total performance impact is indicated by the Impact Score and it is obtained by subtracting the total score obtained by the product out of 190. So the lower the Impact Score, the better. Microsoft comes in last place here with 27.7 points as it obtained the highest Impact Score. In case you are wondering how the AVC score was derived, AV-Comparatives explains: For the AV-C score, we took the rounded mean values of first and subsequent runs for File Copying, whilst for Launching Applications we considered only the subsequent runs. “Very fast” gets 15 points, “fast” gets 10 points, “mediocre” gets 5 points and “slow” gets 0 points. Finally, we have the certifications and as expected, Microsoft received the Standard rating here, which is the lowest rank. It wasn't alone though as Total Defense also received the same certificate as it was just a little bit ahead of Defender in the Impact Score (table in the image above). You can find AV-Comparatives' full test report here.

Beware: Windows 11-ready CPUs with VAES "susceptible to data damage", full CPU list here by Sayan Sen When Windows 11 was first announced, there was a lot of annoyance surrounding it as Microsoft introduced stringent system requirements for it. Only modern microprocessors from AMD and Intel were supported as they were said to possess enhanced security support compared to previous generation CPUs, namely Intel 7th Gen "Kaby Lake" and AMD Zen (Ryzen 1000) or older. However, Microsoft has found that there are issues on the supported CPUs which come with Vectorized AES (VAES) instruction. The company notes that such Windows 11 and Windows Server 2022 devices are "susceptible to data damage" as the Advanced Encryption Standard (AES) instruction is meant to accelerate data encryption and any bugs in this are bound to adversely impact device data. Here is how Microsoft describes the issue: Windows devices that support the newest Vector Advanced Encryption Standard (AES) (VAES) instruction set might be susceptible to data damage. The affected Windows devices use one of the following on new hardware: AES XEX-based tweaked-codebook mode with ciphertext stealing (AES-XTS) AES with Galois/Counter Mode (GCM) (AES-GCM) The good news is that Microsoft has resolved the issue via previous Windows Updates KB5014746 and KB5014019. Users are expected to have performance impact on BitLocker, TLS, and also disk throughput upon installing the workaround updates. The company says that users can expect up to a two times (2x) slow down in AES performance. To prevent further data damage, we addressed this issue in the May 24, 2022 preview release and the June 14, 2022 security release. After applying those updates, you might notice slower performance for almost one month after you install them on Windows Server 2022 and Windows 11 (original release). The scenarios that might have performance degradation include: BitLocker Transport Layer Security (TLS) (specifically load balancers) Disk throughput, especially for enterprise customers Symptoms AES-based operations might be two times (2x) slower after installing the Windows update for the May 24, 2022 preview release or the June 14, 2022 security release. You can find more details on Microsoft's official website where the issue is described under support article KB5017259. Microsoft has not provided an official list of CPUs that are affected, so we did a bit of digging around for our own. The VAES instruction was introduced in 2018, which means all the Windows 11-supported processor models are certainly impacted by this issue. From deep within the interwebs, we discovered that Intel CPUs starting from the 10th Gen Ice Lake 10nm mobile chips are affected as they introduced the VAES instructions for the first time with their new Sunny Cove design. This was a big move from Intel as it was finally moving over from the 2015's Skylake architecture and its iterations. Over on the AMD side, the Zen 3-based Ryzen 5000 series desktop SKUs as well as Ryzen 5000 mobile parts are hit. Interestingly, this isn't the first time that users have faced performance issues with supported Windows 11 processors. Last year, it was revealed that Virtualization-based Security (VBS) was causing a crippling impact in games even on supported chips. Update: Added more CPU families impacted by the issue: You can navigate through all the impacted Ice Lake CPUs on using this tag here. Likewise you can also find the other Zen 3-based CPUs using this tag here. Zen CPUs with 3D V-cache are also affected. These include the Ryzen 7 5800X3D and the EPYC Milan-X CPUs.

Report: Microsoft Defender is hogging Intel CPUs while AMD Ryzen remains unscathed by Sayan Sen AV-Comparatives, an anti-malware assessment company, published a report back in May regarding the performance impact that anti-virus programs can have on a system. In its report, AV-Comparatives found that Microsoft Defender was one of the worst system hoggers and as such, it was only able to secure the "STANDARD" award in the test. Fellow outlet TechPowerUp (TPU) also discovered something similar but more digging has revealed that the performance impact may have to do with a bug in Defender which adversely affects Intel processors while AMD CPUs seem to be unaffected. TPU found that the MsMpEng exe file, which is the anti-malware service process for Microsoft Defender, eats up Intel CPU cycles, thus affecting the performance. The Cinebench R23 rendering benchmark, which is a heavily multi-threaded test, was used to ascertain this behavior. This behavior is perhaps similar to what AV-Comparatives observed as the tested system in that case was also an Intel system, an i3 one to be precise. Three identical test runs were made in two scenarios, one with real-time protection enabled (Bad) and the other with real-time protection disabled (Good). It is seen that the Intel Comet Lake-S Core i9-10850K loses around 6% performance which is quite a lot for such a powerful CPU. Seeing this, it is not hard to imagine why Defender scored so poorly in AV-Comparatives' test with an i3, something which is far less capable than the i9 here. Using Microsoft Sysinternals' Process Explorer feature, this observation was confirmed and looked at more closely. TPU believes that a bug in Defender is causing it to use up more of Intel's Performance Counter Monitors (PCMs) than is necessary, which in turn is causing it to have conflicts with other processes. Intel PCM is basically responsible for measuring internal resource utilization for code executions in Core and Xeon processors. The site says that its new Counter Control software can mitigate this bug which may be used as a temporary workaround until Microsoft releases an official fix for this issue. If you, however, are not affected or bothered by the impact then you can always choose to skip any workaround. Source and images: TechPowerUp

AV-Comparatives' latest test finds Microsoft Defender hogs your system real bad by Sayan Sen It's probably fair to say that Microsoft's Defender hasn't had the best of times recently. A couple of days ago there were reports of Defender for Endpoint causing various issues on client Windows 10 systems. And now, there is bit more bad news as Microsoft's in-house anti-malware product can really hit lower end Windows systems bad according to the latest Performance Impact testing by AV-Comparatives. In the final Awards rating, Defender was barely able to secure the "Standard" rating as it came in second-last in the evaluation alongside Total Defense Anti-Virus. In all, the following anti-malware products were tested: Avast Free Antivirus 22.3 AVG Free Antivirus 22.3 Avira Prime 1.1 Bitdefender Internet Security 26.0 ESET Internet Security 15.1 G Data Total Security 25.5 K7 Total Security 16.0 Kaspersky Internet Security 21.3 Malwarebytes Premium 4.5 McAfee Total Protection 25.5 Microsoft Defender 4.18 NortonLifeLock Norton 360 Deluxe 22.22 Panda Free Antivirus 21.01 TotalAV Antivirus Pro 5.16 Total Defense Essential Antivirus 13.0 Trend Micro Internet Security 17.7 VIPRE Advanced Security 11.0 AV-Comparatives final Awards The following real-world tests were done using an up-to-date Windows 10 21H2 64-bit system with Intel Core-i3, 4GB of RAM, and SSD. The i3 and 4GB RAM was used to simulate typical lower-end PCs which generally are impacted most by anti-virus programs. File copying Archiving / unarchiving Installing / uninstalling applications - using silent install mode Launching applications - Microsoft Office (Word, Excel, PowerPoint) and Adobe Acrobat Reader Downloading files Browsing Websites - using Google Chrome The total score received in the above tests is being referred to as "AV-C Score". Other than the real world tests listed above, the PC Mark 10 Professional Testing Suite synthetic benchmark was also run. Here is how all the products have performed in the tests. The image on the left shows the AV-C performances while the image on the right shows the total scores which also includes the PC Mark scores: If you are wondering what the "Impact Score" is, the column basically represents how far off the total obtained score is from the full marks of 190. Therefore, bigger the Impact Score, the greater performance impact an anti-malware program had on the tested system. For example, if we take Microsoft Defender, it has an Impact Score of 24.6, which implies it has scored 24.6 points less than the full score of 190, ie, 165.4. You can read the original report on AV-Comparatives' site here. If you are wondering how Defender and the other products has done in case of general protection, you can read this article here.