Ram frequency is a joke?

[GSquadron]

I already did some benches myself with my ram from 800 to 1600(O.C) and I didn't see any difference to tell the truth.
Now it comes out that ram frequency is a waste of money from this video

What do you think? I personally believe it

 

[Fourstaff]

Only certain programs are affected by memory, most aren't memory sensitive. This is especially true for Intel's newer platform. If you want to see the difference clearly take a look at APU's performance with different sticks.

 

[RCoon]

I already did some benches myself with my ram from 800 to 1600(O.C) and I didn't see any difference to tell the truth.
Now it comes out that ram frequency is a waste of money from this video

What do you think? I personally believe it

Most people already know this, however RAM frequency IS NOT a joke when it comes to APU integrated graphics - everybody knows that too.

 

[Frick]

800Mhz to 1600Mhz and you didn't see a difference in benches? Exactly what did you benchmark? CPU speed?

But yeah no news here.

 

[GSquadron]

I benched Games only

Why apus? I don't know about that

 

[RCoon]

I benched Games only

Why apus? I don't know about that

APU's utilise system RAM as their VRAM. The higher the frequency, the faster the data can be accessed.

http://hexus.net/tech/reviews/cpu/46073-amd-a10-5800k-trinity-needs-faster-ram/?page=3

 

[CounterZeus]

I bought 16GB of the cheapest 1333MHz CAS9 RAM I could find two years ago for exactly this reason (80 euros then)

 

[THE_EGG]

Yeh frequency of RAM is still important with integrated graphics/apus. There was a test done with 1333, 1600, 1866 and 2133mhz ram sets in a magazine article I read about a week ago. With dedicated graphics there was maybe a 1 or 2 fps difference between them in games (within margin of error) but integrated graphics showed quite significant differences.

 

[vega22]

ram speeds also help a lot in things which are more cpu intensive.

gaming is gpu intensive....so it is not going to have a massive impact is it?

i tested the theory when sandy came out and going from 1333 to 2133 would only get me about 2 fps.

that being said i would rather have 2 more fps than 2 less.

edit

but in reply to your op title, no not ram speeds, it is the headline grabbing journalism which is the joke :thumb:

 

[LAN_deRf_HA]

The benefits are there for games, just not every game. Considering there's been a $10 difference between 1333 and 2133 kits for months now this debate is no longer necessary.

 

[Nordic]

Recent anandtech memory speed scaling test on haswell... here.

Speed also depends on platform.

 

[Sasqui]

It really DOES make a difference. I saw close to a 1 second drop in SuperPI going from 1600 to 2133. This is a real application, look at the difference! From Anand's review...

 

[Ikaruga]

APU's utilise system RAM as their VRAM. The higher the frequency, the faster the data can be accessed.

http://hexus.net/tech/reviews/cpu/46073-amd-a10-5800k-trinity-needs-faster-ram/?page=3

Scaling stops after a certain point tho. Anand and Tom did reviews about this recently.

BTW, the guy is a clueless attention-whore (regardless if he is right or wrong about a particular subject). For examlpe, once he posted a video somewhere (on reddit iirc) where he literally copy pasted pictures from the wikipedia page about anisotropic filtering, but did not read any of it actually (and didn't know what it was also), but (and this is not a joke!) he made up a complete meaningless story about what he thought anisotropic filtering could be, (just like countless other times on other subjects), and he confidently told his "convincing story" to the children of the interwebz, and earned credit for it. When we pointed out that he talks complete bullshit again, he just said sorry, removed the video and earned even more credit with his next (finally correct) take....

I can understand why one might be an attention whore on the internet, but I do not understand why he keeps getting the attention (and new hardware) when he doesn't have an original thought or he is wrong when he does. Just look at how he touches that RAM-modul when he supposed to be the master and teacher of the wannabies who want to be a "pro enthusiast" like him.

tl&dr: idontwanttoliveonthisplanetanymore.jpg

 

[EarthDog]

Outside of a few rare things, ram frequency is a joke, yes. You want to buy ddr3 1600-1866 CL9 1.5v and call it a day. Outside of rare applications like winzip above, iGPU, and multi-gpu setups, anything faster than that will only show gains in benchmarking and is not worth the extra few dollars.

Sasqui did a great job of cherry picking a result though.

Here is single GPU statements from that article:

When using our HD6950 however there seems to be little difference between memory settings with no trend.

Dirt 3 seems to prefer setups with a lower CL – MHz does not seem to have any effect.

Single GPU frame rates for Bioshock has no direct effect for memory changes with less than 2% covering our range of tests.

Again, we see no obvious trend in average frame rates for a discrete GPU.

Frame rates for Sleeping Dogs vary between 49.3 FPS and 49.6 FPS, showing no distinct improvement for certain memory timings.

Also, that testing was done at a VERY low resolution which puts more stress on the CPU which will exacerbate the results. things will likely change at 1920x1080 res where it is more GPU bound than CPU showing even less difference.

 

[Sasqui]

Sasqui did a great job of cherry picking a result though.

First one that my attention whore self could find, that supported my completely biased opinion.

Some apps it doesn't make much of a difference, but there are some that it really does. So yes, faster RAM is better!!!

 

[EarthDog]

So yes, faster RAM is better!!!

If you use those rare apps, sure. But a blanket statement like that is not true as it doesn't help in most cases.

 

[Sasqui]

If you use those rare apps, sure. But a blanket statement like that is not true as it doesn't help in most cases.

How can you say that higher memory bandwidth won't move data faster? It's the whole game changer with the IMC, and what made AMD stand out before intel got into the same game.

Charts here are rife with examples: http://www.anandtech.com/show/7364/memory-scaling-on-haswell/6

 

[Frick]

How can you say that higher memory bandwidth won't move data faster? It's the whole game changer with the IMC, and what made AMD stand out before intel got into the same game.

Charts here are rife with examples: http://www.anandtech.com/show/7364/memory-scaling-on-haswell/6

Oooooh, sometimes you actually gain an entire frame per second! Amazing!

No wait, look at this:

That is nearing two!

 

[TheLaughingMan]

FPS is not where faster RAM is going to help in a game. The difference will show in level loading times. Obviously in situations where the RAM is being shared as VRAM for an APU or other intergrate solution it can make a huge difference in graphical performance.

Now RAM speed can also have a big impact on data intensive tasks like video encoding and archiving programs like 7-Zip. Note the change in performance for the Intel chip with just the change in RAM speed.

http://www.futurelooks.com/amds-new-vishera-fx-8350-am3-cpu-reviewed/4/

 

[Sasqui]

Oooooh, sometimes you actually gain an entire frame per second! Amazing!

See, that's 10 frikking precent improvement! Yeah!!!

 

[Frick]

That's 10 frikking precent improvement! Yeah!!!

And neither of them tangible.

 

[Sasqui]

And neither of them tangible.

I'll take that tangible any day.

 

[EarthDog]

FPS is not where faster RAM is going to help in a game. The difference will show in level loading times.

It would? I doubt that as a HDD or SSD is exponentially slower than ram... Would you happen to have any links proving that assertion?

How can you say that higher memory bandwidth won't move data faster? It's the whole game changer with the IMC, and what made AMD stand out before intel got into the same game.

Charts here are rife with examples: http://www.anandtech.com/show/7364/memory-scaling-on-haswell/6

How about an analogy for ya...

Let's say you have 1 GPM max flow from your spicket at your house. Would it make any difference in the flow rate if you have a garden hose or a fire hose hooked up to it? No because you are limited to the flow rate of your spicket. In the case of computing, the bandwidth isn't the bottle neck in most cases/applications, so why would having more bandwidth available make a difference if the data going through it isn't saturating the pipe?

 

[Sasqui]

Let's say you have 1 GPM max flow from your spicket at your house. Would it make any difference in the flow rate if you have a garden hose or a fire hose hooked up to it? No because you are limited to the flow rate of your spicket. In the case of computing, the bandwidth isn't the bottle neck in most cases/applications, so why would having more bandwidth available make a difference if the data going through it isn't saturating the pipe?

You're talking bottlenecks now. There are clearly applications that gain from faster bandwidth, those that have a lot of CPU -> Memory I/O

AutoCAD is a great example. First load, it grabs all program code from the HDD and puts into memory... slow. Close AutoCAD... it keeps a sizeable chunk of code in memory for subsequent loads. The second time you load AutoCAD, it's roughly 1/4 of the time to load. Most code is coming from memory, not the HDD. Faster memory = faster load the 2nd time.

With the logic you're using, an SSD wouldn't be any faster than a HDD.

 

[red_stapler]

I'd be interested to see if the AI / simulation speed scales with RAM speed on a game like Sins of a Solar Empire, or Supreme Commander.