Dual-channel DDR3 really faster than Dual-channel DDR2??

  • Thread starter Thread starter Man-wai Chang
  • Start date Start date
M

Man-wai Chang

By how much? Would one feel a difference?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.36
^ ^ 18:28:01 up 12 days 23:36 1 user load average: 0.01 0.01 0.00
¤£­É¶U! ¤£¶BÄF! ¤£´©¥æ! ¤£¥´¥æ! ¤£¥´§T! ¤£¦Û±þ! ½Ð¦Ò¼{ºî´© (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa
 
By how much? Would one feel a difference?

I asked this question back in 2009:
http://www.howtofixcomputers.com/fo...iple-channel-vs-ddr2-dual-channel-243403.html

Have things changed?

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.36
^ ^ 16:42:01 up 13 days 21:50 1 user load average: 0.12 0.06 0.06
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa
 
Man-wai Chang said:

I don't get the impression the bus protocols are any different.

When if comes to memory controller design, the controllers
don't use all the state machine features of newer memories.
So the behavior with DDR2 and DDR3, could be largely
the same.

And that makes it a bit easier to compare them. If you
want additional transfer bandwidth, then simply compare
the clock rates, to understand which is faster. Or, if you
think an application is sensitive to memory latency, take
the CAS number, times the nanoseconds per memory clock tick,
to work out what the latency would be. CAS10 at DDR2-800
would be the same latency as CAS5 at DDR2-400.

I would say, sure, DDR3 is faster. You can get DDR3-2000
DIMMs for example, which gives a tremendous transfer bandwidth.
Now, how often does that bandwidth make a difference ? Not
that often. It might be most apparent, if you were benchmarking
WINRAR at identical processor clock speeds. It's most likely
to be important when doing synthetic benchmarks. But if you
expected your desktop to be "more snappy", I doubt you could
see a difference that way.

My old Core2 Duo system only managed about 3GB/sec of memory
bandwidth, and yet that system was responsive. My current
system isn't much faster on the memory bus, but I really
can't tell the difference. The improvement in memory
performance, doesn't seem to matter much here.

*******

You can run your own benchmark series, to answer questions
like this. If your BIOS has options for memory speed,
such as DDR2-533, DDR2-667, DDR2-800, try testing with
all three, using something like SuperPI as a benchmark.
Do you see much of a difference in the benchmark. If
you had DDR3 memory, the difference might be, it allows
a higher clock rate, like the above mentioned DDR3-2000
type sticks. And your chipset has to support going that
fast, to make it worthwhile buying sticks like that.

Paul
 
You can run your own benchmark series, to answer questions
like this. If your BIOS has options for memory speed,
such as DDR2-533, DDR2-667, DDR2-800, try testing with
all three, using something like SuperPI as a benchmark.
Do you see much of a difference in the benchmark. If
you had DDR3 memory, the difference might be, it allows
a higher clock rate, like the above mentioned DDR3-2000
type sticks. And your chipset has to support going that
fast, to make it worthwhile buying sticks like that.

I guess my PC would stay with me for a while...

--
@~@ Might, Courage, Vision, SINCERITY.
/ v \ Simplicity is Beauty! May the Force and Farce be with you!
/( _ )\ (x86_64 Ubuntu 9.10) Linux 2.6.36
^ ^ 21:12:01 up 15 days 2:20 1 user load average: 0.09 0.06 0.01
ä¸å€Ÿè²¸! ä¸è©é¨™! ä¸æ´äº¤! ä¸æ‰“交! ä¸æ‰“劫! ä¸è‡ªæ®º! è«‹è€ƒæ…®ç¶œæ´ (CSSA):
http://www.swd.gov.hk/tc/index/site_pubsvc/page_socsecu/sub_addressesa
 
Back
Top