Sign in with
Sign up | Sign in
Your question

Mixing RAM of different speeds

  • Memory
  • DDR2
  • RAM
  • Hewlett Packard
Last response: in Memory
a b } Memory
a b α HP
February 17, 2011 6:38:55 AM

I have 2.5GB of DDR2 memory, a 2GB module and a 512MB module. It's an HP that originally had 2x512MB=1GB of DDR2 667.

The 2GB module is a Trancend DDR2 800MHz "JetRAM"
The 512MB module is a (insert cheapest brand name in the world) DDR2 667 MHz "HP-F**kRAM"

I heard that the 800MHz RAM has to run at the same speed as the slower ram, is this true?
Is my 800MHz RAM running at 667MHz or can they both run at their own speeds?

Should I remove the 512MB 667MHz module?

More about : mixing ram speeds

a c 347 } Memory
February 17, 2011 1:44:12 PM

The 'LCD' {Lowest Common Denominator} is in effect, and the BIOS 'should' be manually set if possible to accommodate the 'mixing'. However, the biggest problems are the Voltage and then the CAS. Generally, the BIOS reads the SPD info and sets the speed and timings to a safe number, and if the BIOS cannot it often Disables the DIMM slot(s) to avoid errors.

800 MHz 5-5-5-15 2.0v **
667 MHz 4-4-4-16 2.3v
667 MHz 5-5-5-16 2.15~2.2v

**If you run CPUz the 800 MHz RAM probably has a lower CAS timing set when running at 667 MHz; example 4-4-4-14 and might too have a lower SPD voltage requirement.

Example {values from CPUz}:
667 MHz 4-4-4-14 2.0v **
667 MHz 4-4-4-16 2.3v
667 MHz 4-4-4-16 2.15~2.2v

Please keep in mind the above information is hypothetical and doesn't represent your current RAM. As far as 'slowing', unknown it might go either way, but in all probability a differences that would require a bench test like Super PI to detect. If the 800 MHz {faster} has higher CAS {slower} vs 667 MHz {slower} has lower CAS {faster} the next affect lowering BOTH Frequency and CAS might cancel each other out. I see 1333 MHz DDR3 CAS 8 beating 1600 MHz CAS 9 frequently; the true speed is the quality of the RAM IC.

Test, when mixing RAM it is a good idea to run Memtest86+ to make certain it's not producing errors. I recommend a minimum of 2 passes and 4 passes is preferred.

CPUz -
Memtest86+ - use the ISO/zip and burn a bootable CD/DVD.
Super PI -

CPUz SPD tab

a b } Memory
a b α HP
February 17, 2011 3:38:31 PM

I've been running it like this for over 2 years now so I doubt there are any errors.

About Super PI,
The ONLY thing that I don't know about computers is about RAM, I have only a rough idea about RAM latency and timings so what exactly do each of those numbers mean in brief, and what should I do with super PI?
Related resources
a c 347 } Memory
February 17, 2011 4:22:13 PM

Unfortunately, CAS and Frequency alone don't guarantee which RAM is faster - the quality of the IC is what truly determines the overall performance. Also, the tightness of the CAS is good indicator; example 9-9-9-24 is often faster than 8-9-8-24. The true test is a benching applications like Super PI which is a very RAM intensive free application.

If you're curious about your 'mix' then test with Super PI, next pull the 512MB stick and retest, next pull the 2GB and test with 2X512MB. Compare the results. Anything else is a subjective 'guessing game.'

My assumption is the results will be small.

Comparison Module #1: Mushkin Redline 996805 6-8-6-24 1600MHz
Comparison Module #2: G.Skill Pi Series 7-8-7-24 1600MHz
Comparison Module #3: Mushkin 998687 8-9-8-24 1866MHz
Comparison Module #4: Corsair Dominator TR3X6G1600C8D 8-8-8-24 1600MHz
Comparison Module #5: Kingston HyperX T1 9-9-9-27 1600MHz
Comparison Module #6: Mushkin Blackline 998677B 7-7-7-20 1333MHz
Comparison Module #7: Patriot Viper II Sector 7 9-9-9-27 1800MHz


Overclocked all at 1800 MHz:

a b } Memory
a b α HP
February 17, 2011 4:24:17 PM

Ok, I'll do that ASAP,
but can you give me a brief explanation of what RAM timings, latency and CAS are and how they relate to frequency?
a c 347 } Memory
February 17, 2011 5:34:52 PM

That's more of an engineering and physics question. CAS Timings are inversely related to Frequency, and the timings are related to and limited by the Sine wave nodal openings a/k/a Address Strobe, the IC and the RAM's ability to receive and transmit data. The quality of the IC both determines the Frequency, CAS, and other factors like leakage, etc. It's lithography similar to a CPU determines performance: CAS, Storage, Thermal characteristics, Type, etc.

Beyond that Benchmarks are the true predictor of measurable performance.

"Secondly, the advantages of higher clock speeds are in part negated by memory latency, since memory access times have not been able to keep pace with increasing clock frequencies."

So from my point of view - the most efficient 'fast' RAM is in comparison to having high frequencies with lower tight CAS timings. Example: 1600 MHz 8-8-8-24, is typically faster than 1600 MHz 9-9-9-24 or 1600 MHz 8-9-8-24. Further, 1333 MHz 7-7-7-24 can be faster than 1600 9-9-9-24. And no - adding up all the numbers doesn't guarantee the results either.

edit: found this chart: