1600mhz cl7 vs 2400mhz cl11 with AMD FX 8350 at 5ghz

[jackalopeater]

Hey guys,

 

I've done all kinds of testing with the FX 8350 from clock speeds to core configurations but I've never thought to test memory settings.  I've always read and have been told that with AMD cpu's Latency is king.  But no one really explained that to me.  Sure I could find the "equations" that should give me an idea of the performance delta.

 

What really got me thinking about it was I found a pair of G.Skill Trident X's on Newegg for really cheap, but they were 1600mhz cl7.  I currently have a pair of AMD Radeon R9 Gamer Ram at 2400mhz cl11.

 

 

 

I figured this would be a perfect scenario to test timings at least.  I know there can be some discrepancies caused by the tertiary timings since I did not mess with those.

 

System Specs:

AMD FX 8350 5ghz

Asus ROG Crosshair V Formula Z

AMD Radeon R9 Gamer Ram 2400mhz

XFX R9 295x2

Cooler Master HAF Stacker

Cooler Master V1200p

Swiftech h320

 

All that being said, I ran one set of tests at 2400mhz with 11-12-12-31 timings then again at 1600mhz with 7-8-8-24

 

Here are my results

 

Cinebench and Aida64 cache & memory test

 

1600mhz 

http://i.imgur.com/Bg4ufOv.jpg

 

2400mhz

http://i.imgur.com/9IeI77m.jpg

 

 

Firestrike 

 

1600mhz

http://i.imgur.com/91opxjT.jpg

 

 

2400mhz

http://i.imgur.com/qoil9fD.jpg

 

 

ROG RealBench

 

1600mhz

http://i.imgur.com/sP9BXFZ.jpg

 

2400mhz

http://i.imgur.com/ecfPCJg.jpg

 

 

SuperPi 

 

1600mhz

http://i.imgur.com/OwbnGy4.jpg

 

2400mhz

http://i.imgur.com/9H5zUHC.jpg

 

WinRar

 

1600mhz

http://i.imgur.com/5TNiXhm.jpg

 

2400mhz

http://i.imgur.com/nSnZp9u.jpg

 

So far my results tell me one thing.....You can get really good results from lower speed with much tighter timings.

The importance of this is because not all AMD motherboards will support 2400mhz ram, and not all AMD cpu's can handle 2400mhz ram (after all they're only rated up to 1866mhz)

 

So if you're not willing to risk it for the uncertainty of the faster ram, because some people aren't, some may be better off with quality lower speed and tighter timing ram.

 

 

If there is anything I could add or you would like to know, please feel free to ask

[CoolaxGaming]

Hmm...

Isint it proven already that ram speeds beyond 1333 do minimal difference?

[Enderman]

It only makes a difference in APUs because the integrated graphics uses your system ram

 

the 8350 does not have an iGPU

those few numbers difference in your benchmarks make 0 difference in the real world

you will see 0 fps difference between 1600 and 3200MHz during games and 0 seconds faster in programs

 

spending more money for higher frequency is absolutely not worth any additional cost

[ProKoN]

nice testing

 

run the tests 3-5 times each and take the average score, dont just run it once and call it a day.

 

good call on the 2400mhz. i could never get 2400mhz to run with an 8350 i had in the past

 

the tertiary timings do change a fair bit when scaling the frequency up, they can impact performance.

 

run the winrar and maxxmem benchmarks.

 

also the aida64 memory benchmark is a good one.

[GoldSrc]

Those are synthetic benchmarks, which mean pretty much nothing in the real world.

Real-world performance between 1333 and 2400MHz is pretty much nothing. 

[Lays]

Hmm...

Isint it proven already that ram speeds beyond 1333 do minimal difference?

 

 

Completely depends on what you're doing, super CPU intensive games will benefit from faster ram & tighter timings, as well as benchmarks.

[jackalopeater]

Okay, I'll be doing some tests with extractions as well as maybe some video rendering times with CPU only.

 

I'm not entirely versed in Memory Overclocking and tweaking, so I'm trying to nail down some results and collect them in a place.

 

Feedback is always welcome, especially when it's construct.

 

So far,so good guys! Keep it rolling

[-rascal-]

2400 MHz is pretty impressive to get stable on a FX-8350 @ 5GHz...but the mix of 11 and 12 DRAM timings with 1.65V makes it a let down. Loosen the timings, and push a bunch of voltage into it -- not that hard to get 2400 MHz to run.

 

If you can get the 2400MHz kit down to CL9, 10, or what-not, that is definitely something.

 

IMO, I would stick with the G.Skill 1600MHz CL7 kit.

 

 

Off-topic: Where did you get dat shiny AMD decal? 

[Jumper118]

It only makes a difference in APUs because the integrated graphics uses your system ram

 

the 8350 does not have an iGPU

those few numbers difference in your benchmarks make 0 difference in the real world

you will see 0 fps difference between 1600 and 3200MHz during games and 0 seconds faster in programs

 

spending more money for higher frequency is absolutely not worth any additional cost

actually the ram speed can make your boot up time noticeably faster. 

Hey guys,

 

I've done all kinds of testing with the FX 8350 from clock speeds to core configurations but I've never thought to test memory settings.  I've always read and have been told that with AMD cpu's Latency is king.  But no one really explained that to me.  Sure I could find the "equations" that should give me an idea of the performance delta.

 

What really got me thinking about it was I found a pair of G.Skill Trident X's on Newegg for really cheap, but they were 1600mhz cl7.  I currently have a pair of AMD Radeon R9 Gamer Ram at 2400mhz cl11.

 

 

 

 

I figured this would be a perfect scenario to test timings at least.  I know there can be some discrepancies caused by the tertiary timings since I did not mess with those.

 

System Specs:

AMD FX 8350 5ghz

Asus ROG Crosshair V Formula Z

AMD Radeon R9 Gamer Ram 2400mhz

XFX R9 295x2

Cooler Master HAF Stacker

Cooler Master V1200p

Swiftech h320

 

 

 

If there is anything I could add or you would like to know, please feel free to ask

if you increase the nb and ht link frequency more then it should make more use of the memory bandwidth as i think AIDA64 shows slightly lower results for the 1600mhz, but that might be because the nb frequency is lower. 

[jackalopeater]

2400 MHz is pretty impressive to get stable on a FX-8350 @ 5GHz...but the mix of 11 and 12 DRAM timings with 1.65V makes it a let down. Loosen the timings, and push a bunch of voltage into it -- not that hard to get 2400 MHz to run.

 

If you can get the 2400MHz kit down to CL9, 10, or what-not, that is definitely something.

 

IMO, I would stick with the G.Skill 1600MHz CL7 kit.

 

 

Off-topic: Where did you get dat shiny AMD decal? 

 

Yeah, I ran the 2400mhz at it's stock AMP profile, but I got the decal from @westickitout on twitter

 

 

actually the ram speed can make your boot up time noticeably faster. 

if you increase the nb and ht link frequency more then it should make more use of the memory bandwidth as i think AIDA64 shows slightly lower results for the 1600mhz, but that might be because the nb frequency is lower. 

 

 

Yeah, I'm sure increasing the nb and ht link will.  Not sure why Aida64 derped out on the 1600mhz readings, but it was all at stock nb and ht link speeds for both tests.  The only overclocking I did was purely Multiplier base

[Jumper118]

Yeah, I ran the 2400mhz at it's stock AMP profile, but I got the decal from @westickitout on twitter

 

 

Yeah, I'm sure increasing the nb and ht link will.  Not sure why Aida64 derped out on the 1600mhz readings, but it was all at stock nb and ht link speeds for both tests.  The only overclocking I did was purely Multiplier base

it wasn't at stock the nb was 2200 for the 1600mhz test and 2400mhz for the 2400mhz test, you should test both ram speeds with the 2400mhz nb for it to be fair

[jackalopeater]

it wasn't at stock the nb was 2200 for the 1600mhz test and 2400mhz for the 2400mhz test, you should test both ram speeds with the 2400mhz nb for it to be fair

Whoa! I didn't even catch that, not sure what changed it in the UEFI since I didn't.....Time to retest while manually setting that!

 

Big thanks for catch!

 

I'll be rerunning the tests and editing the OP.  

 

now how do I get you a cookie

[Jumper118]

Whoa! I didn't even catch that, not sure what changed it in the UEFI since I didn't.....Time to retest while manually setting that!

 

Big thanks for catch!

good. it should make a difference. 

[jackalopeater]

OP updated to reflect NB being at 2400mhz for both speeds