CAS vs Speed

scote7877

Junior Member
Jan 27, 2006
10
0
0
I am thinking about getting some RAM. As an example, you can get 2 Gb of DDR2 667 CAS3 for about the same price as 2 GB of DDR2 800 CAS4. As a gamer, which would be better; the higher CAS or the higher MHz, assuming stock speeds?
 

SerpentRoyal

Banned
May 20, 2007
3,517
0
0
Core speed is KING. I'd stay with high quality DDR2 800 RAMs rated for 1.8V operation. These sticks should be able to hit 450MHz with 2.0-2.1V.
 

hans007

Lifer
Feb 1, 2000
20,212
18
81
actually if you ran them at stock cl3 667 is faster than 800 cl4. and on an intel box, it doesnt even really matter how much mhz of ram you have as long as you fill the core 2 duo bus.

i.e. if your cpu is a 1333 bus, you need 2 667 modules to maximize its bandwidth.

latency is in cycles. a 667 mhz cycle is 1sec/667 or 1.5 ns a 800mhz cycle is then 1.2ns. thus CL3 @ 667 is 3 cycles of 1.5 = 4.5 for an access. cl4 @ 800 is 1.2 ns x 4 cycles = 4.8 ns for an access. so yeah serpent royal is wrong about it being all about core speed.


800mhz @ 1.8v is the jedec standard. all ram that is rated for more than 1.8ns at a standard speed (i.e. 667 , 800) is actually out of spec. usually if you buy "valueram" which is designed to run on machines like HPs and dells that only take jedec standard ram, it will run at 1.8. so getting cl5 800 1.8v "valueram" such as kingston, is usually the most economical sinec you can easily get them to cl4 or lower with the voltage (thus you are overvoltaging and running out of jedec spec, but if you are overclocking this really was the whole point anyway)

 

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
Originally posted by: hans007
actually if you ran them at stock cl3 667 is faster than 800 cl4. and on an intel box, it doesnt even really matter how much mhz of ram you have as long as you fill the core 2 duo bus.

i.e. if your cpu is a 1333 bus, you need 2 667 modules to maximize its bandwidth.

latency is in cycles. a 667 mhz cycle is 1sec/667 or 1.5 ns a 800mhz cycle is then 1.2ns. thus CL3 @ 667 is 3 cycles of 1.5 = 4.5 for an access. cl4 @ 800 is 1.2 ns x 4 cycles = 4.8 ns for an access. so yeah serpent royal is wrong about it being all about core speed.


800mhz @ 1.8v is the jedec standard. all ram that is rated for more than 1.8ns at a standard speed (i.e. 667 , 800) is actually out of spec. usually if you buy "valueram" which is designed to run on machines like HPs and dells that only take jedec standard ram, it will run at 1.8. so getting cl5 800 1.8v "valueram" such as kingston, is usually the most economical sinec you can easily get them to cl4 or lower with the voltage (thus you are overvoltaging and running out of jedec spec, but if you are overclocking this really was the whole point anyway)

read the link I posted...very detailed explination with a chart. Your numbers are incorrect BTW...check the link.
 

BitByBit

Senior member
Jan 2, 2005
474
2
81
Originally posted by: hans007
latency is in cycles. a 667 mhz cycle is 1sec/667 or 1.5 ns a 800mhz cycle is then 1.2ns. thus CL3 @ 667 is 3 cycles of 1.5 = 4.5 for an access. cl4 @ 800 is 1.2 ns x 4 cycles = 4.8 ns for an access. so yeah serpent royal is wrong about it being all about core speed.

DDR2-667 actually operates at an internal clock of 166MHz. Its CAS Latency is measured (confusingly) in external clocks, which is always half of the effective clock for double data-rate RAM. So a CL3 DDR2-667 module has a latency of:

3 / 333MHz = 9ns.

 

JustaGeek

Platinum Member
Jan 27, 2007
2,827
0
71
Nnnnnooo... the memory bus operates at 333MHz, for the DDR memory speed of 667MHZ.

You just multiply it by 2 for Double Data Rate.

Only the CPU FSB is Quad pumped, e.g. 266 x 4 = 1066MHz.
 

BitByBit

Senior member
Jan 2, 2005
474
2
81
Originally posted by: JustaGeek
Nnnnnooo... the memory bus operates at 333MHz, for the DDR memory speed of 667MHZ.

You just multiply it by 2 for Double Data Rate.

Only the CPU FSB is Quad pumped, e.g. 266 x 4 = 1066MHz.

Yyyeesss...
DDR2 operates at an internal clock speed of one quarter its rated speed, and one half the external, or memory bus.
See here.

The reason for DDR2's increased latency in external clocks is the halved internal frequency. With DDR, the internal frequency was the same as the external, which meant that a one cycle delay internally became one externally. With DDR2, one internal cycle becomes two externally.
DDR3 takes that further, with an internal clock of 1/8 the rated speed, and 1/4 of the external bus.
 

JustaGeek

Platinum Member
Jan 27, 2007
2,827
0
71
Yup, it seems that you are right. I copied the excerpts from that article here:

"Like all SDRAM implementations, DDR2 stores memory in memory cells that are activated with the use of a clock signal to synchronize their operation with an external data bus. Like DDR before it, DDR2 cells transfer data both on the rising and falling edge of the clock (a technique called double pumping). The key difference between DDR and DDR2 is that in DDR2 the bus is clocked at twice the speed of the memory cells, so four words of data can be transferred per memory cell cycle. Thus, without speeding up the memory cells themselves, DDR2 can effectively operate at twice the bus speed of DDR."

The memory divider actually raises the 2x memory bus speed, e.g. 266MHz x 1.5 = 400MHz, which is already twice the memory bus speed of 200MHz.

It makes it confusing as hell for everyone, but based on the above it is correct.

Thanks, BitByBit.