Linked by Thom Holwerda on Wed 12th Nov 2008 22:55 UTC
Editorial Yesterday, a story made its rounds across the internet. It was picked up by many large news websites, and I'm sure it will be quoted by people until eternity. It was published by a large website, looked all fancy, it had multiple pages - it looked like it was really something. However, anyone with even the remotest bit of knowledge knows that the article was a collection of complete and utter bogus.
Permalink for comment 337203
To read all comments associated with this story, please click here.
Pretty Hilarious
by PlatformAgnostic on Thu 13th Nov 2008 11:40 UTC
Member since:

Here's why you should discount everything this guy says: "The additional Local Procedure Call overhead of moving portions of MDAC (Microsoft Data Access Components) out of the kernel would most certainly be felt by a time-sensitive, looping transactional workload like ADO Stress. "

It seems like he can't tell the difference between the Kernel and random user-mode DLLs that have nothing to do with the kernel at all. I can safely say that MDAC has not moved up, down, into, or out of the kernel, since it's a relatively high-layer user mode component. Also, a change like the one he mentioned would have a much greater than 5% overhead.

There's a lot of tuning, tweaking, and performance work that goes on toward the end of the design cycle. There's really no point in trying to draw conclusious from memory and CPU data from the M3 build. There have been quite a few changes since 6801 to the kernel, heap, and other subsystems that would affect this dramatically. I bet the thread-count hasn't changed much though, so maybe that "benchmark" will stand.

I can't believe this guy is still hawking his benchmark scripts...

Edited 2008-11-13 11:41 UTC

Reply Score: 7