VOGONS


First post, by 386SX

User metadata
Rank l33t
Rank
l33t

Hi,
trying to push the Barton 3200+/KT600 to its limits I already bough 2x1Gb Corsair CL3 ram before the Sapphire HD3450 AGP that being a 2009 card should help to "support/accelerate" Office/2D/Web usage up its limits. I was wondering it switching from the 3.3.3.8 memory to 2.3.2.5 Kit I could buy, will result in some speed differences. Ok that the faster the better but did you feel the change in similar "timing only" switch?
Thank

Last edited by 386SX on 2016-10-29, 15:52. Edited 1 time in total.

Reply 1 of 6, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

The difference will show up in benchmarks, but will be small. Going with nForce can also give you more performance.

Check the specifications of the memory, often high speed memory of that era needs extra voltage and manual settings of timings. The default SPD settings will be conservative like 3-3-3-8.

YouTube, Facebook, Website

Reply 2 of 6, by 386SX

User metadata
Rank l33t
Rank
l33t
PhilsComputerLab wrote:

The difference will show up in benchmarks, but will be small. Going with nForce can also give you more performance.

Check the specifications of the memory, often high speed memory of that era needs extra voltage and manual settings of timings. The default SPD settings will be conservative like 3-3-3-8.

The ram is Corsair VS1GB400C3 and the specs says 3-3-3-8 at 2,5v. Strange thing is that in ubuntu with the decode dimm command it says for one 3-3-3-8 and for the other 3-4-4-8 when they are same model with same chips (even if one is a bit taller than the other like manufactured in another version.

Reply 4 of 6, by 386SX

User metadata
Rank l33t
Rank
l33t
PhilsComputerLab wrote:

Yea that's standard RAM. You need to look for DDR400 CL2 memory.

Thanks, I'll try to push a bit more speed from this system, actually the fastest I can build with the hardware I got. 😁

Reply 5 of 6, by Stermy57

User metadata
Rank Newbie
Rank
Newbie
386SX wrote:
PhilsComputerLab wrote:

The difference will show up in benchmarks, but will be small. Going with nForce can also give you more performance.

Check the specifications of the memory, often high speed memory of that era needs extra voltage and manual settings of timings. The default SPD settings will be conservative like 3-3-3-8.

The ram is Corsair VS1GB400C3 and the specs says 3-3-3-8 at 2,5v. Strange thing is that in ubuntu with the decode dimm command it says for one 3-3-3-8 and for the other 3-4-4-8 when they are same model with same chips (even if one is a bit taller than the other like manufactured in another version.

Corsair Value Select were popular, they used different type of IC not always the same (the cheapest available)
Lot of them used Samsung UCCC famous to running at 250mhz 3-3-3-8 1T at default voltages... IC DDR1 made by Samsung don't like voltage; I know this because I have 2 kit of these particular Corsair Ram.
Anyway you can push them only with 754 or 939 platform.
With 462 or 478 the best options are winbond BH-5/BH-6 (256*2), UTT are in general more difficult to get stable.

Reply 6 of 6, by shamino

User metadata
Rank l33t
Rank
l33t

I remember on the nForce2 there was a strange quirk where it was actually slower if the last latency figure was set too low (I don't know what that number represents). So 2-2-2-5 would benchmark slower, I think optimal on that chipset was 2-2-2-11 or something like that. I have some Kingston HyperX which I think is speced for 2-3-2-6 but on my nForce2 I always ran it at 2-3-2-11 for that reason (that's from memory, could be wrong on the details).
I have no idea about VIA, maybe on the KT600 the fastest possible settings would perform as they should, but if you get fast RAM for that board I'd try experimenting with the settings just in case it has any such quirks.