At 12:47 PM 8/29/98 -0500, you wrote:
At 10:58 AM 8/29/98 -0400, you wrote:
Actually my experience with Shuttle has been pretty good. The real issue with the TX chip set is it wont cache memeory over 64MB. So if you expand pass that, you take a BIG performance penalty.
That is incorrect. The performance hit is small (<5%) and it evens out becuase you access the hard drive less frequently.
Where do you get this number ( < 5 %)? I would think that to measure the performance of a computer having no cache above 64 M one would have to do really specific tests and the result would be very application dependent. Your post makes it sound as if caching has not much effect at all. I think that's giving the wrong picture. You say, it "evens out..." - maybe under certain circumstances with heavy disk I/O and using DMA but on the other side, when you have a CPU-intensive application using more than 64 M or a multiuser system with more than one user you most certainly pay a penalty. I personally would stay away from a system like that as well as other Intel marketing gimmicks like Celeron. Samartha PS.: On the Intel web site, there is a comparison of Celeron processor with the same speed, one with and one without cache. - To get out of this list, please send email to majordomo@suse.com with this text in its body: unsubscribe suse-linux-e