Introducing the USB-C 20G Loopback Plugs! Validate USB 3.2 Gen 2x2 ports with 20Gb/s testing performance. Compact, reliable, and built for production testing, diagnostics, and high-speed port verification. For more information or to order visit: https://t.co/J6p5W1EO9F
Version V11.7 of MemTest86 was released today.
With support for the LPCAMM2 RAM form factor, including detecting which chip is bad. Plus a bunch of other minor bug fixes and improvements.
https://t.co/v9Np57f0ys
@lafaiel@x86deadandback Integer test result only contributes a small amount of the overall CPUmark score. It is interesting from a historical point of point, allowing comparison to old CPUs. Yes, this sub-test is synthetic, but the instruction mix was from real code. Integer maths is pretty common.
@TheProfosist@vxunderground No, that was a separate problem. Had a DoS attack on web site over the week-end. Peaked at around 4000 requests a second (normal traffic is 100/s). Should be sorted out now.
@shanselman@vxunderground Scott we also got our account blocked without notice. This is for MemTest86 development, which Microsoft use internally for testing their Surface hardware. So if you do find the reason, we'd be very interested to know.
There has been around 5 significant benchmark algorithms updates over the last 20 years. Including moving to 64bit, using AES hardware & adding SIMD to some of the sub-tests. Overall result is scaled across releases to allow a rough comparison. While there has been significant improvement in a few narrow areas (like AES encryption) overall single threaded general compute has processed only slowly.
This also (partially) explains why new computer don't feel dramatically faster and old computers don't feel slow, in general desktop usage.
Introducing the USB-C Short Circuit Tester! Compact, practical, and delivers rapid testing of port voltages and short circuits— under 1 second per port — perfect for fast production line checks.
For more information or to order visit: https://t.co/VZQ7OXYw10
@Olrak29_ With a sensible naming scheme there would a lot less need to visit review & benchmark sites. So we fully support this naming madness.
https://t.co/kcsb1cReS9
Pulling apart USB Type-C cables today trying to work out why some models have errors at 20Gb and some don't. We suspect the (lack of) twist in the data lines near the termination give this one occasional errors.