Access latency to main memory

Access latency to main memory

Hi,

I'm working on a memory analysis tools, and I was wondering if it is possible to have a 'general estimation' of the latency for accessing the main memory. By general estimation I mean that I don't need 'exact' values, but only a rule-of-thumb for knowing (estimating) the access penalty to main memory. An good example would be a typical (average) cycle number.

These are the specification of the system i'm working on:

Intel Core 2 Quad Q9950 @2.83 GHz
DIMM DDR2 Synchronous 800 MHz (64 bit)

Hopefully my question is not too vague. Thank you in advance.

2 posts / 0 nouveau(x)
Dernière contribution
Reportez-vous à notre Notice d'optimisation pour plus d'informations sur les choix et l'optimisation des performances dans les produits logiciels Intel.

the access latency may be very different depending on the architecture of your processor and memory type. Also on multi-socket systems with NUMA (Non-Uniform Memory Access) the distance to the processor (number of hops)with accessed memory plays a big role (latency grows as measured in ns or in cycles). There are open source benchmark kits that can measure memory latency on your particular system (one of them is LMbench).

Best wishes,
Roman

Laisser un commentaire

Veuillez ouvrir une session pour ajouter un commentaire. Pas encore membre ? Rejoignez-nous dès aujourd’hui