Is Intel screwing us over

[Sakkura]

No it's not.

You cannot compare the different speed RAMs as that changes the variable which we are trying to measure. We aren't measuring the impact of RAM frequency on performance, we are measuring the impact of single vs dual channel.

You clearly don't have an understanding of the situation, and it's clear to me that I've been arguing with a troll.

 

Yes it is. And what it does is it demonstrates that the performance continues to scale with memory bandwidth long past the point we're discussing. It's a well established fact that there are diminishing returns on memory performance, which means that proving a significant gain all the way up to quad channel DDR3-2400 proves beyond a shadow of a doubt that there's going to be a significant gain from single channel to dual channel. It's absolutely ridiculous to claim that there would be no difference from single to dual channel, and then suddenly a difference as you move to quad channel and faster memory. That you turn around from such absurd claims to call me a troll without understanding of the situation is just astonishing.

[stconquest]

Yes it is. And what it does is it demonstrates that the performance continues to scale with memory bandwidth long past the point we're discussing. It's a well established fact that there are diminishing returns on memory performance, which means that proving a significant gain all the way up to quad channel DDR3-2400 proves beyond a shadow of a doubt that there's going to be a significant gain from single channel to dual channel. It's absolutely ridiculous to claim that there would be no difference from single to dual channel, and then suddenly a difference as you move to quad channel and faster memory. That you turn around from such absurd claims to call me a troll without understanding of the situation is just astonishing.

 

What is saying is that you can't compare dual channel 1600Mhz RAM against quad channel 2133Mhz and argue that it is the extra bandwidth is what is causing the performance difference.  You now have two variables that affect the results, not just one.

[TheMcSame]

Look! Facts! Figures! 

 

http://www.tomshardware.com/reviews/PARALLEL-PROCESSING,1705-11.html

http://www.reddit.com/r/buildapc/comments/1fcs77/discussion_ram_single_vs_dual_channel_speed/

http://www.anandtech.com/show/8672/lenovo-thinkstation-p300-workstation-review-haswell-plus-quadro/6

 

 

I followed this thread to see what people had to say about "intel screwing us over", not some people arguing over something as stupid as dual channel vs single channel. 

[Sakkura]

What is saying is that you can't compare dual channel 1600Mhz RAM against quad channel 2133Mhz and argue that it is the extra bandwidth is what is causing the performance difference.  You now have two variables that affect the results, not just one.

 

But going from dual to quad channel at the same clocks still increases performance, and then scaling to higher clocks increases it even more. So that demonstrates that the game scales very well with memory bandwidth.

[stconquest]

But going from dual to quad channel at the same clocks still increases performance, and then scaling to higher clocks increases it even more. So that demonstrates that the game scales very well with memory bandwidth.

 

Not 130-150...  144-150.

[Sakkura]

Not 130-150...  144-150.

 

But you go from 130 to 150 by adding memory bandwidth beyond a dual-channel configuration. So it's ridiculous to claim that going to single to dual channel will make no difference.

[stconquest]

But you go from 130 to 150 by adding memory bandwidth beyond a dual-channel configuration. So it's ridiculous to claim that going to single to dual channel will make no difference.

 

No, your graph showed the difference between 1600Mhz dual channel up to 2133 quad channel in order to achieve that 20 point difference.    There are two factors at play, the bandwidth AND the clock.

 

[Faceman]

No, your graph showed the difference between 1600Mhz dual channel up to 2133 quad channel in order to achieve that 20 point difference.    There are two factors at play, the bandwidth AND the clock.

 

You can't with this guy, he doesn't get it.

[Nena Trinity]

No, your graph showed the difference between 1600Mhz dual channel up to 2133 quad channel in order to achieve that 20 point difference.    There are two factors at play, the bandwidth AND the clock.

 

You two still on this? Ram barely does anything unless you have APU!

[Sakkura]

No, your graph showed the difference between 1600Mhz dual channel up to 2133 quad channel in order to achieve that 20 point difference.    There are two factors at play, the bandwidth AND the clock.

 

 

The graph showed a difference from 2133 dual channel to 2133 quad channel as well. The point is, from standard DDR3-1600 dual channel there's still AT LEAST another 15% performance to be gained simply from a memory upgrade. There would be a lot more to gain if you were coming from DDR3-1600 single channel, because there are diminishing returns.

 

You two still on this? Ram barely does anything unless you have APU!

 

It does something in certain tasks, like compression, transcoding, and a few particularly memory-sensitive games.

[stconquest]

"The graph showed a difference from 2133 dual channel to 2133 quad channel as well. The point is, from standard DDR3-1600 dual channel there's still AT LEAST another 15% performance to be gained simply from a memory upgrade. There would be a lot more to gain if you were coming from DDR3-1600 single channel, because there are diminishing returns."

 

 

Not 130-150...  144-150.

 

Full loop.

 

 

So, back on topic here...

[stconquest]

I don't believe it is just Intel screwing over the consumers, it is not that cut and dry.  You can't simply blame one company that uses the system to a sort of advantage.  Sure they could push for changing how things work... but they are on top, so changing it means risk.   That could mean lost revenue.

 

I was serious about releasing the x86 and x64 architectures.  Sorry for my lack of fully understanding.  These laws exist to induce competition, but having two companies "competing" in this important field is proving to not be driving innovation.  It seems to be inefficient in a field that is closely related to the math sciences.  In math class you either get the answer right, or wrong... there are no marks for "good effort".  If there is a mathematically better solution, all the parties involved (tech companies, legislative bodies) are holding it back.... to make money. Sad, but not unique by any means.

[Nena Trinity]

The graph showed a difference from 2133 dual channel to 2133 quad channel as well. The point is, from standard DDR3-1600 dual channel there's still AT LEAST another 15% performance to be gained simply from a memory upgrade. There would be a lot more to gain if you were coming from DDR3-1600 single channel, because there are diminishing returns.

 

 

It does something in certain tasks, like compression, transcoding, and a few particularly memory-sensitive games.

True but for regular use and gaming nothing.

 

If I rendered like 10 videos a week I would need more RAM!