dmccunney composed on 2021-03-11 09:51 (UTC-0500): > Felix Miata wrote:
>> Odds are that 32GB capable board features dual channel RAM. >> https://en.wikipedia.org/wiki/Multi-channel_memory_architecture > Possible. >> IME when RAM is not used in matched pairs in correct slots in a dual channel >> board, RAM speed (memtest86) is cut by nearly half. Did you test RAM speed >> before >> and after the change? > No. I simply made sure I had RAM that matched the specs of the other > sticks. The only difference was that one stick is 8GB instead of > four. I was *not* using RAM of different speeds, and no mismatch was > involved.. > I saw *no* negative performance impact, and would have been startled if I did. By not matching size of pairs, you disable dual channel. You should run memtest86 with and without the 4G and 8G sticks to see the difference in print on your screen. A quick test here using MemTest86 V8.3 Free on: CPU: AMD A10-7850K Radeon R7, 3.7Ghz motherboard: ASUSTeK model: A88X-PRO RAM: Mushkin DDR3-2133 XMP, 10-12-12-28, 2 sticks of 4GB each matched pair RAM speed: 7474 MB/s dual channel single stick RAM speed: 5943 MB/s not dual channel = 79.5% CPU: Intel Pentium G4600, 3.6GHz motherboard: ASUSTeK model: B85M-E RAM: Crucial DDR4-2400, 17-17-17-39, 2 sticks of 8GB each matched pair RAM speed: 21.67 GB/s dual channel single stick RAM speed: 13.92 GB/s not dual channel = 64.2% RAM: generic DDR4-2400, 17-17-17-39, 2 sticks of 8GB each matched pair RAM speed: 19.92 GB/s dual channel single stick RAM speed: 12.07 GB/s not dual channel = 60.6% My recollection is with DDR2 the difference tended to be bigger, as low as 53% for 1 stick compared to dual channel. -- Evolution as taught in public schools, like religion, is based on faith, not on science. Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/ _______________________________________________ Freedos-user mailing list Freedos-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/freedos-user
