2400 cl10 or 1600 cl7 for same price.

[EvilDen]

Hello everybody.
Help me to chose, which ram kit to choose.
Two options:G skill F3-1600C7D-16GTX vs F3-2400C10D-16GTX. Both kits are nearly for same price at this moment.
I have i5 3570k and asus v gene.
I mainly use my pc for simple tasks, but sometimes like to play games as Batman Arkham origins and Far Cry 3. At least 20 hours weekly. Which kit will give better performance? I am worried a bit about 2400mhz kit voltage-1.65. Is it safe to use under xmp profile?

 

[MeteorsRaining]

Technically, higher the MHz/CL ratio, the better.
Therefore 2400/10=240.
And 1600/7=~228.

1600/7 is more than enough for games. Voltage is not a problem the MoBo will adjust it so the 2400 will eventually come down in speed. And both will be roughly same.

Get the 1600/7.

 

[EvilDen]

Thanks for help.
About adjusting the voltage by motherboard...Does it happen always or can manually change/control in bios?

 

[MeteorsRaining]

We can manually change the profile but it happens automatically in most MoBos (the downscale of MHz). 2400MHz is of no use in gaming, its for CPU intensive tasks like heavy video rendering/ VFX/ etc.

 

[EvilDen]

We can manually change the profile but it happens automatically in most MoBos (the downscale of MHz). 2400MHz is of no use in gaming, its for CPU intensive tasks like heavy video rendering/ VFX/ etc.

Thank you.
1600mhz should be safer as well, it uses just 1.5 v. Besides kit has spd - 1600 Mhz with cl 9, better then standard Jedec 1333.
I have no doubts now, thanks to you. Have chatted with g skill support team, they didn`t give me any proper advice, just generalized.

 

[MeteorsRaining]

Glad to help

 

[Pinhedd]

Technically, higher the MHz/CL ratio, the better.
Therefore 2400/10=240.
And 1600/7=~228.

Huh? There's no technical basis for that at all, it's just something that you or someone else made up.

 

[anbello262]

I'd certainly suggest the 2400 cl10, if they're the same price... Clearly better, and cl10@2400 is about 8.33 nanoseconds, while cl7@1600 is 8.75 nanoseconds.
Therefore, the latency on the 1600cl7 is actually worse, as well as the frecuency. The 2400 is better on both aspects.

 

[Pinhedd]

I'd certainly suggest the 2400 cl10, if they're the same price... Clearly better, and cl10@2400 is about 8.33 nanoseconds, while cl7@1600 is 8.75 nanoseconds.
Therefore, the latency on the 1600cl7 is actually worse, as well as the frecuency. The 2400 is better on both aspects.

the latency is largely immaterial. It's just a number that is used by the memory controller to schedule read and write operations. The impact on performance is almost negligible. The mean access time for a random DRAM read on most modern Intel microprocessors is around 53ns + about 30 cycles of CPU overhead. IO bus bandwidth is much, much more important.

 

[MeteorsRaining]

You certainly say that latency doesn't matter?

I disagree, you won't enjoy a 1600/11 RAM. Moreover, Speed doesn't really matter in games, it does matter, but largely in rendering/ 3D developments/ VFX/ etc where the speed decides the overall performance.
See his usage, he's gonna use the PC for simple tasks/ games where speed don't really matter.

So its just a matter of better CL, which'll decide how fast the RAM accesses a cloumn in the module and get it to the memory controller from the time it's requested by the controller.

Also, the 2400 RAM has 1.65V which is not really a point to worry, but the performance will be downscaled automatically to get the voltage to 1.5V. So what's the real difference in buying a higher MHz RAM (the 2400) which will get downscaled and may even perform worse than 1600MHz RAM in games due to Voltage?

IMO, 1600/8, 1866/9, 2133/10 and 2400/11 are all the same for gaming.

Contrary to the most popular of beliefs, memory kits do not work as stated out of the box. The number of times I have walked through a large LAN event and found people playing games on $2000+ water cooled systems, only to find that their kit of DDR3-2400 is actually running at DDR3-1333 astounds me. It is a lot more common than you think, and there is probably someone you know that is a culprit of this. Making sure memory is set at its rated speed is an important part of the process, and as an enthusiast we have a job to make sure that is the case.

Taken from: http://www.anandtech.com/show/6372/memory-performance-16gb-ddr31333-to-ddr32400-on-ivy-bridge-igp-with-gskill

More reference:
http://www.tomshardware.com/answers/id-2040165/1600-mhz-cl7-2400-mhz-cl10.html

And, the MoBo won't take more than 1600MHz for Haswell refresh, you'd need to enable XMP and manually set the speed in BIOS if you wish to use 2400MHz. 1600MHz will run at defualt w/o such issues.

What I'm trying to establish is not that 2400/10 is not good, it sure is, but it has almost negligible effect on gaming vs a 1600/7 RAM.

Other reference:
https://www.youtube.com/watch?v=dWgzA2C61z4

 

[Pinhedd]

You certainly say that latency doesn't matter?

I disagree, you won't enjoy a 1600/11 RAM. Moreover, Speed doesn't really matter in games, it does matter, but largely in rendering/ 3D developments/ VFX/ etc where the speed decides the overall performance.
See his usage, he's gonna use the PC for simple tasks/ games where speed don't really matter.

So its just a matter of better CL, which'll decide how fast the RAM accesses a cloumn in the module and get it to the memory controller from the time it's requested by the controller.

Also, the 2400 RAM has 1.65V which is not really a point to worry, but the performance will be downscaled automatically to get the voltage to 1.5V. So what's the real difference in buying a higher MHz RAM (the 2400) which will get downscaled and may even perform worse than 1600MHz RAM in games due to Voltage?

IMO, 1600/8, 1866/9, 2133/10 and 2400/11 are all the same for gaming.

Contrary to the most popular of beliefs, memory kits do not work as stated out of the box. The number of times I have walked through a large LAN event and found people playing games on $2000+ water cooled systems, only to find that their kit of DDR3-2400 is actually running at DDR3-1333 astounds me. It is a lot more common than you think, and there is probably someone you know that is a culprit of this. Making sure memory is set at its rated speed is an important part of the process, and as an enthusiast we have a job to make sure that is the case.

Taken from: http://www.anandtech.com/show/6372/memory-performance-16gb-ddr31333-to-ddr32400-on-ivy-bridge-igp-with-gskill

More reference:
http://www.tomshardware.com/answers/id-2040165/1600-mhz-cl7-2400-mhz-cl10.html

And, the MoBo won't take more than 1600MHz for Haswell refresh, you'd need to enable XMP and manually set the speed in BIOS if you wish to use 2400MHz. 1600MHz will run at defualt w/o such issues.

What I'm trying to establish is not that 2400/10 is not good, it sure is, but it has almost negligible effect on gaming vs a 1600/7 RAM.

Other reference:
https://www.youtube.com/watch?v=dWgzA2C61z4

Your Anandtech source (not that Anandtech is the greatest source, I've found many glaring flaws in their articles) quite clearly reinforces exactly what I just wrote, that the data transfer rate is substantially more important than the column access latency. The performance gain relative to DDR3-1333 using DDR3-2400 is on average twice as high as using DDR3-1600. Go ahead and play around with various column latency values and see what kind of spreads are obtained.

A good memory controller can easily mask the effects of column latency at high data rates, a poor memory controller cannot; Intel's DDR3 memory controller is considered to be very, very good. Contrary to popular belief the objective of the DRAM bus is not to obtain the necessary data as fast as possible, it is to transfer as much data as possible and keep the bus as busy as is needed.

Also, I'm not sure where you're getting information about memory voltage from. DRAM ICs are all designed to operate at 1.5 volts which is the JEDEC standard. Samples that meet particular performance metrics may be binned as DDR3L models which will operate stably when configured with either a 1.5 volt or 1.35 volt supply, while others may be binned as high performance models which will operate at particularly tight non-standard timings and high data rates when configured with a supply above 1.5 volts. DDR3-2400 memory is simply memory that has been tested to run at DDR3-2400 data rates under certain operating conditions, it can be configured using a large number of different configurations, including many which may not be officially supported by the manufacturer; it is the responsibility of the system operator to enable the features which configure the memory in a fashion that is not universally supported. Giving memory extra voltage to work with, tightening the timings, and increasing the data rate may have a slight impact on performance but the user does so at his or her own risk.

 

[MeteorsRaining]

But my friend, I'm not saying that 2400MHz RAM is not good. What I'm trying to establish is that a 2400MHz is not really ideal for gaming if he doesn't want to involve himself in changing BIOS settings, and let me be clear, I'm in full support of playing around with the BIOS for good. Its just that 1600/7 will do the job, performance wise, and won't require him to change anything anywhere, just plug and play sort of.

Yes I absolutely agree that speed does matter, but I'm yet to see someone complaining even with a 1600/8 RAM about bad performance while gaming, let alone 1600/7. What does one really need 2400 for? Highly CPU intensive tasks one would say, and gaming isn't one of them, 3D dev/ VFX/ rendering/ desgining/ etc are, there, the higher the speed the better.

Yes the 2400/10 is an awesome RAM, but 1600/7 is what he'll need for his tasks. I'm a bit unsure of what you're really standing for, 2400 as a better RAM or the point that Speed matters more than CL. I can only say 1600/7 is more than enough for gaming and all general tasks, 2400/10 is not what I'd get.

Also, I'm a VFX Developer and desgining university student and have literally seen and worked on every RAM from 800MHz/CL5 to 2666MHz/CL11. To be on the topic, I've 2400/10 in my own rig which is for VFX tasks and gaming too. I see alot of difference while using VFX softwares on my rig and the other on I have at my work place in project development area which has 1866/9 RAM. But I can scarcly make out any difference while gaming (the one at work has i5 2500k and doesn't really have any good games, I've put some for tp 😀 ) like minmal difference in FPS while on BF3 on High, both give off 80+ FPS, that too because my personal rig has superior CPU and other specs. But I know that's not the way to compare.

Numerous other workstations there have from 1333MHz to 2666MHz modules and from Pentium 4 to Sandy-E so I'm a bit well experienced there. I may be wrong, but what I tell is from my knowledge and more importantly first hand experience. I agree with the Voltage part completely.

 

[Pinhedd]

But my friend, I'm not saying that 2400MHz RAM is not good. What I'm trying to establish is that a 2400MHz is not really ideal for gaming if he doesn't want to involve himself in changing BIOS settings, and let me be clear, I'm in full support of playing around with the BIOS for good. Its just that 1600/7 will do the job, performance wise, and won't require him to change anything anywhere, just plug and play sort of.

Yes I absolutely agree that speed does matter, but I'm yet to see someone complaining even with a 1600/8 RAM about bad performance while gaming, let alone 1600/7. What does one really need 2400 for? Highly CPU intensive tasks one would say, and gaming isn't one of them, 3D dev/ VFX/ rendering/ desgining/ etc are, there, the higher the speed the better.

Yes the 2400/10 is an awesome RAM, but 1600/7 is what he'll need for his tasks. I'm a bit unsure of what you're really standing for, 2400 as a better RAM or the point that Speed matters more than CL. I can only say 1600/7 is more than enough for gaming and all general tasks, 2400/10 is not what I'd get.

Also, I'm a VFX Developer and desgining university student and have literally seen and worked on every RAM from 800MHz/CL5 to 2666MHz/CL11. To be on the topic, I've 2400/10 in my own rig which is for VFX tasks and gaming too. I see alot of difference while using VFX softwares on my rig and the other on I have at my work place in project development area which has 1866/9 RAM. But I can scarcly make out any difference while gaming (the one at work has i5 2500k and doesn't really have any good games, I've put some for tp 😀 ) like minmal difference in FPS while on BF3 on High, both give off 80+ FPS, that too because my personal rig has superior CPU and other specs. But I know that's not the way to compare.

Numerous other workstations there have from 1333MHz to 2666MHz modules and from Pentium 4 to Sandy-E so I'm a bit well experienced there. I may be wrong, but what I tell is from my knowledge and more importantly first hand experience. I agree with the Voltage part completely.

It's not your recommendation that I'm taking issue with, because for what it's worth I agree with it. DDR3-1600 is well supported and will present few complications either now or in the future.

What I take issue with is your assertion that there's some sort of technical basis to relative performance being derived from the ratio of data rate to column latency. There is no such thing.

 

[MeteorsRaining]

I agree, actually that's the 'simple way' of sorting RAM out, using the simplest ratio possible. I know its not correct way to go about comparing sticks. I knew I made a mistake by writing 'technically' instead of 'simply put'.

Apologies for jumbling words up, won't happen in future. I know Mhz/CL ratio is not an indicator of performance, but it do tells us that 1600/8 is better than 1600/10 and so on. But then I agree its not an excuse to give it the 'technical' tag. Sorry for that, I didn't want to give out the wrong info, just wrote the wrong word and described it so well (though not correct) if I was in an exam I'd have got A1 lol 😀