Interesting numbers benchmarking RAM PC2-4200 vs PC2-5300 vs PC-6400

I was trying to see if RAM speed made much difference on my CF-51RCLDFBM, especially since I upgraded the cpu to a Core 2 Duo 2.0 GHz (Intel T7200).

I tested four sets of RAM each test was using two 1GB SO-DIMMs

The test were done using the Novabench RAM transfer speed test MB/s.

First test was one Hynix and one Samsung PC2-4200 SO-DIMMs

Results in MB/s 1045, 1044, 1053

Second test was two Samsung PC-5300 SO-DIMMs

Results in MB/s 735, 752, 757

Third test was two Hynix PC2-6400 SO-DIMMs

Results in MB/s 743, 740, 743

Fourth test was two Kingston PC2-4200 SO-DIMMs

Results in MB/s 1192, 1220, 1215

It was interesting that the PC2-4200 memory moved data faster that is why I dug up the second set and tested them. I thought the PC2-5300 would probably be best with the Core 2 Duo upgrade.

 

That seems about right. Kingston is better memory than Hynix. Even though the RAM is slower it will run better because that's what the machine is designed to run at!

It's the same concept as putting premium fuel in a Prius. The prissy prius is designed to take normal, cheap, 87 fuel because it is a slow crappy car. If you put premium in it you will see NO BENEFIT due to he fact that it's a stupid, slow hybrid.

~Rob - suckin gas and haulin @$$!~

 

It may very well be that your MB is the limiting factor; I realize the memory controller is embedded in the T7200, but that works in conjunction with the Northbridge controller to deliver data as well. As your Northbridge is for an older single-core processor, I'd tend to blame it in this case; it may not be able to keep data in sync at the faster bus speeds.

mnem
smoof.

 

Rep to ya mate.

 

Sorry, I don't know English well enough to use jokes correctly.
But it's not true.

Main difference in RAM is latency. Maybe in this case PC2-4200 memory has better latency parameters than PC2-6400. Or chipset doesn't know how to use this faster memory correctly.
Anyway, I'm sure it's impossible to feel those results

 

I agree it takes a significant improvement for you to actually feel the difference. I do a lot of video editing and I don't upgrade until I read where the benchmarks show that my rendering times will be cut in half. Of course rendering times are a little more tangible than an improvement in RAM transfer speed.

I have a good friend that is an admin and he always jokes about "the seat of the pants feel", because it does take a big jump in benchmark numbers for user to feel it.

 

Thats true for sure, but you, like a lot of people here, know how much difference a well tuned/maintained/kept clean system makes when compared with a bloat-ware filled/bad designed/poorly put together system makes.
Just take a new Acer, Packard Bell or the likes and look how long it takes to boot. Now take your Toughbook with the same processor, memory and disk and watch the difference....?!
It's not just bloatware, or a bit better memory or a little faster HDD, efficient MoBo etc. etc. but combined together all these little things can make a big difference!

 

What were the settings for the memory kits? There are several threads comparing memory speed oh identical systems, benchmarks don't really translate to real life performance except in gaming on integrated graphics.

Sent from my Galaxy Nexus using Tapatalk

 

Since the ToughBooks are not really performance computers, I didn't bother with timings since you can't change them in the BIOS like some of the motherboards designed for overclocking. Plus it was not new memory so I didn't have the timings listed on the package. But when I get a chance I can throw them back in and check the timings with some utility like cpuz.

OK, had a little time this morning to get the timings.

Also found a set of Panasonic replacement memory that is PC2-4200 which according to CPU-Z is Samsung memory.

The RAM transfer speed test for the Panasonic marked Samsung PC2-4200 was:

MB/s: 1071, 1061, 1047

So it fell in line with the other PC2-4200.

OK now for timings in this order DRAM Frequency, CAS# Latency, RAS# to CAS# Delay, RAS# Precharge, Cycle time, Bank cycle time.

Mixed Samsung/Hynix PC2-4200
266, 4, 4, 4, 12, 16

Samsung marked as Panasonic PC2-4200
266, 4, 4, 4, 11, 15

Kingston PC2-4200
266, 4, 4, 4, 12, 16

Samsung PC2-5300
333, 5, 5, 5, 15, 21

Hynix PC2-6400
333, 5, 5, 5, 15, 5*

*I think this must be an off reading because the chipset is having trouble with the timings. Because the Timing table list the Bank cycle time as 20 which falls more in line with what you would think.

Again this information was obtained using a CF-51RCLDFBM with an Intel Core 2 Duo 2.0 GHz processor.

 

Novabench - 2533MB/s on my config.

Aida64: