Learn how to a build a cloud-first strategyRegister Now

x
  • Status: Solved
  • Priority: Medium
  • Security: Public
  • Views: 1257
  • Last Modified:

Memory speed vs size?

EE,

I have a Dell R610 that can fit 96GB with 8GB sticks (KTD - PE313/8G)

I'm not sure how to tell whether it's dual or quad rank but my question is that either one when fully populated into the slots will reduce speed from 1333 down to 800hz

My question is, is this a noticable difference in memory performance and what can the drawbacks be? We don't exactly need 96GB per server and can increase the speed to 1333 at 48GB but have no bench testing or examples to confirm a noticable benefit

Please assist, thanks
0
snyderkv
Asked:
snyderkv
3 Solutions
 
airborne1128Commented:
You will not be able to tell the difference, but you will see the bechmark difference.  What is that you hope to gain?
0
 
andyalderSaggar makers bottom knockerCommented:
The chart entitled "Relative Performance for Different Memory Speeds" in http://globalsp.ts.fujitsu.com/dmsp/Publications/public/wp-westmere-ep-memory-performance-ww-en.pdf is very useful; it shows a performance drop from 1.0 to 0.62 when they ran STREAM memory bandwidth sensitive test but only a drop from 1.0 to 0.91 for SPECint. They also deliberately messed up the channel balancing in "Relative Performance for Different Interleaving Levels" graph and again a huge drop in performance using STREAM but far less drop when using the more real world app SPECint test.
0
 
CallandorCommented:
Do you have Xeon 5500 series cpus?  The Xeon 5500 series uses a an onboard memory controller and QPI transport for communication between processors.  This document on p.16 says 1066 memory has 20-25% higher throughput than 800 memory, and 1333 memory has 3.5-9% higher throughput than 1066 memory: ftp://ftp.hp.com/pub/c-products/servers/options/Memory-Config-Recommendations-for-Intel-Xeon-5500-Series-Servers-Rev1.pdf
0
[Video] Oticon Case Study

Open office environments can create the dynamics for innovation, but they also bring some challenges. With over 1,000 employees in an open office, Oticon needed a solution that would preserve the environment while mitigating disruptive background noises.

Watch how they did it.

 
andyalderSaggar makers bottom knockerCommented:
Just like the white papers on Intel's site the HP white paper doesn't address the speed difference using a real world application but uses a memory performance test. Generally applications perform better if they have more RAM to play with regardless of the extra RAM possibly slowing the bus down.
0
 
snyderkvAuthor Commented:
Cool thanks for the info and the links. I'll check them out

Funny though, the BIOS says 1333 @ 96G despite what the documentation says. I'm guessing it's running slower but displaying it's rated output.

Thanks
0
 
snyderkvAuthor Commented:
Thanks Andy. But I mean we use maybe <10% of the RAM available so I'm sure there's RAM sitting idle?
0
 
andyalderSaggar makers bottom knockerCommented:
BIOS often reports DIMMs ratings, depends on server model. If you're not going to be using the RAM then better to take it out, as noted it does slow it down if fitted but not used.
0
 
snyderkvAuthor Commented:
Roger
0

Featured Post

How to Use the Help Bell

Need to boost the visibility of your question for solutions? Use the Experts Exchange Help Bell to confirm priority levels and contact subject-matter experts for question attention.  Check out this how-to article for more information.

Tackle projects and never again get stuck behind a technical roadblock.
Join Now