Skip to content
TheChatArchiveThe Chat Archive
Perplexitysonar-reasoning-pro-high· PerplexityHugging Face LMArena Search Arena 24K

DDR5 different models read/write speeds (in actual Mbits/s as of some benchmark)…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-search-preview-high’s answer.
  1. User

    DDR5 different models read/write speeds (in actual Mbits/s as of some benchmark) summarized in a table. Show the table's units not in Mbits/sec or similar to it, bit in MHz in a reasonable way.

  2. Perplexity Assistant

    Based on the provided search results, here's a summary of DDR5 RAM read/write/copy speeds from benchmarks:

    DDR5 Memory Performance Benchmarks

    DDR5 Speed RatingConfigurationREAD (MB/s)WRITE (MB/s)COPY (MB/s)Latency (ns)
    4800 MT/s4 sticks, 96GB69,01968,48269,81676.93
    4800 MT/s2 sticks, 64GB71,03371,58372,05877.70
    5600 MT/sG.Skill Trident Z5 Neo87,196Not specifiedNot specified51.50
    6000 MT/s2 sticks, 64GB87,34385,59185,53570.43
    6200 MT/s2 sticks, 64GB90,26888,71488,17968.57
    7600 MT/sPatriot Viper Xtreme 5 RGB110,000+96,000Not specifiedNot specified

    Notes:

    • The DDR5 speed rating (e.g., 4800 MT/s) refers to the memory's rated transfer speed in Mega Transfers per second
    • Actual bandwidth measurements (READ/WRITE/COPY) are from AIDA64 benchmarks in MB/s
    • The Patriot Viper Xtreme 5 RGB kit recorded the highest speeds in TechRadar testing at over 110,000 MB/s read speeds[7]
    • There's a strong linear correlation between memory speed rating and actual performance in benchmarks[1]
    • Performance varies based on configuration (number of sticks and ranks)[9]
    • Single-rank configurations (1Rx8) typically have lower copy speeds compared to dual-rank configurations[9]
    • Using 4 sticks (2DPC) typically reduces performance compared to 2 sticks (1DPC)[9]

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.