desktop processors

Winsage
September 27, 2024
Microsoft is revisiting its Recall feature for Copilot+ on Windows PCs after facing backlash over security issues during its initial rollout. The original Recall captured screenshots and text using optical character recognition (OCR) and stored this data unencrypted, raising security concerns. The feature was postponed before its planned launch on June 18 to address these issues. The revised Recall will be opt-in by default, requiring user activation, and will necessitate re-authentication through Windows Hello to access data. All locally stored Recall data will be encrypted, with encryption keys stored in the device's Trusted Platform Module (TPM). Recall will only function when BitLocker or Device Encryption is enabled, along with Virtualization-Based Security (VBS) and Hypervisor-Protected Code Integrity (HVCI). Recall will operate within a VBS enclave, isolating data in memory and allowing access only after user permission. Only Microsoft-signed code can execute within these enclaves, enhancing security against malware. Users must re-authenticate with Windows Hello each time they access Recall, with biometric authentication as the primary method. The encryption keys are tied to the user's identity, addressing vulnerabilities from the original version. Users can configure settings to limit data collection, manage disk space, and filter sensitive content. Recall will only be available on new PCs meeting specific requirements, including at least 16GB of RAM, 256GB of storage, and a compatible neural processing unit (NPU). Eligible devices include select Arm Windows PCs with Snapdragon chips and x86 PCs with Intel's Core Ultra or AMD's Ryzen AI processors. Microsoft has not provided a timeline for the reintroduction of Recall but indicated it would roll out to Windows Insiders in October.
Winsage
September 26, 2024
AMD's Zen 5 desktop processors, specifically the Ryzen 9000 series, have received a lukewarm reception despite offering improved speed over the previous generation. The Ryzen 9 9950X is noted for its multi-threaded performance, while the Ryzen 7 9700X and Ryzen 5 9600X are considered solid mainstream options. The Ryzen 7 7700X is priced below 0, while the 9700X is priced at 0 but does not deliver a corresponding performance increase. Testing revealed discrepancies in performance claims, attributed to AMD's use of a 'full admin' account during testing. Following the KB5041587 Windows update and BIOS adjustments, a retest showed negligible performance differences in common CPU benchmarks. The testing setup included a fresh installation of Windows 11 and adherence to AMD's protocols. Core parking was enabled for dual CCD processors, but its necessity for Zen 5 chips remains unclear. The Ryzen 5 9600X and Ryzen 7 9700X showed minimal gains in gaming performance under a 105 W TDP limit, with significant improvements only in heavy multi-threaded applications. Dual CCD processors, such as the Ryzen 9 9900X and 9950X, did not receive a new power limit but benefited from the Windows update, although branch prediction optimizations did not impact gaming performance. Core parking negatively affected performance in some titles, and the Windows update did not yield significant improvements for content creation or multi-threaded benchmarks. Users on the older AM4 platform are awaiting 3D V-Cache variants, while those on AM5 are satisfied with current performance. Future releases, like the Ryzen 7 9800X3D, may show only marginal gains.
Winsage
September 23, 2024
AMD unveiled the Ryzen 9000 series desktop processors, promising a 16% increase in instructions per clock (IPC). However, third-party reviews indicated that performance gains were not consistent, with AMD acknowledging potential bugs related to Windows Account privileges. Essential branch prediction optimizations were missing in Windows 11, affecting performance compared to Linux. A Microsoft patch improved performance by about 10% in gaming scenarios, but results varied between installations. AMD also released firmware increasing the thermal design power (TDP) from 65 watts to 105 watts, potentially boosting performance by up to 13%. Additionally, AMD addressed latency issues with the Zen 5 architecture through the AGESA PI 1.2.0.1A firmware update. Tests by Phoronix showed improved performance of the Ryzen 9950X with the updated BIOS on Linux, although overall performance remained stable across 385 benchmarks.
Winsage
September 6, 2024
AMD's Ryzen 9000 series desktop processors have generated mixed performance reviews, with some users reporting modest gains and others noting better improvements. A YouTube analysis by Hardware Unboxed (HwU) found that Windows 11 24H2 showed an average framerate boost of up to 11% compared to Windows 11 23H2 after applying a Ryzen branch prediction optimization patch. However, individual user experiences may vary due to potential underlying issues in the operating system. HwU categorized installations as "Good Install" for better performance and "Bad Install" for inferior results. In their tests, a 9% performance difference was recorded in favor of 24H2 against a "Good Install" of 23H2, while an 11% improvement was noted across 43 titles with a "Bad Install" of 23H2. Factors such as Virtualization-Based Security (VBS) and Hypervisor Code Integrity (HVCI) may contribute to performance inconsistencies.
Winsage
August 31, 2024
Microsoft is integrating deep learning capabilities into Windows 11 with a feature called Windows Recall, which enhances search functionality by periodically capturing screenshots of the user's PC. Users can opt to remove this feature if desired. The rollout of Windows Recall has been delayed due to security and privacy concerns, with plans to introduce it to Windows Insiders in October. An optional Windows update (KB5041865) will allow users to completely remove Recall from their systems. The broader rollout of Windows Recall may occur in early 2025, and it will require devices with a neural processing unit (NPU), such as the upcoming Intel Core Ultra 200 series CPUs, while AMD's Ryzen 9000 processors lack this technology.
Winsage
August 24, 2024
AMD addressed concerns about the performance of its Ryzen 9000 Series desktop processors, attributing subpar results from independent testers to variations in system and software configurations. The company stated that its testing was conducted in Admin mode, which affected the outcomes reported. AMD's tests were performed with Windows Virtualization-based Security (VBS) enabled, which can impact performance, and independent tests showed the Ryzen 9950X was only about three percent faster than the Ryzen 7950X. Additionally, TechSpot reported that the 9950X was six percent slower than Intel's Core i9-14900K in multiple game tests. AMD claimed that the Ryzen 9000 Series offers a 10 percent improvement in productivity and creative workloads, a 25 percent improvement in AI workloads, and a 5-8 percent improvement in gaming compared to the Ryzen 7000 Series. The company announced that optimized branch prediction code for its processors will be available in Windows 11, version 24H2, which is expected to enhance performance. AMD is also working on improving chipset drivers to facilitate easier transitions between CPUs.
Winsage
August 22, 2024
AMD's launch of the Ryzen 9000 desktop processors revealed performance discrepancies between the company's marketing claims and independent reviewer results. AMD identified several factors contributing to these differences, including variations in Windows operating modes, VBS security settings, Intel system configurations, and game selections for benchmarking. The company used an automated testing framework in Super Admin mode, which allowed for optimizations not present in the standard user mode used by reviewers. This led to inflated performance numbers that did not account for the differences between the two modes. AMD's David McAfee stated that the gap in performance between Super Admin and user modes had widened over time, which they had not anticipated. Additionally, the selection of games and specific scenes for benchmarking significantly impacted performance results. McAfee clarified that AMD did not blame reviewers for the discrepancies, attributing them instead to its own testing decisions. AMD is committed to refining its testing methodologies for better alignment with real-world user experiences.
Search