VOGONS


First post, by konc

User metadata
Rank l33t
Rank
l33t

I was always wondering what the actual performance gain from tight memory timings is. OK, we all know that for example a lower CL value means a faster module, but how exactly does this really affect overall performance? Sorry if most of you already know the answer, but this was something still undocumented for me. Well numbers is my kind of thing so I found a 256MB PC133 module rated for CL/RCD/RP = 3/3/3, confirmed with memtest etc that it can tolerate 2/2/2 and ran some benchmarks.

This is not about absolute numbers and performance, but about the difference when only changing memory timings on the same pc.

PCP Player & Quake are from Phil’s DOS benchmarks package and the third is 3D Mark 99 MAX. Test system is a PIII 800EB / Via Apollo 133 / Voodoo 3 2000 / Windows 98SE stock. DOS benchmarks were of course conducted in pure DOS mode.

zmNkYG2l.png

Reply 1 of 3, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

Great job!

I found the same on BX440 machines. At 100 MHz FSB, most, if not all, PC133 modules can handle CL2 and that offers quite a bit of extra performance.

YouTube, Facebook, Website

Reply 2 of 3, by elianda

User metadata
Rank l33t
Rank
l33t

Due to the caches the gain is much lower than on old machines, such as 486/386.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 3 of 3, by noshutdown

User metadata
Rank Oldbie
Rank
Oldbie

this result seems reasonable. on cpus with higher clock ratio like celeron-1.4g, where memory bandwidth is a scarce resource, 2-2-2-5 and 3-3-3-7 memory timing can yield up to 10% difference.