Quote:My memory is supposedly running at 2100 MHz (factory OC'd) but ill have to check on that. (1050 MHz each way if you want to be completely wrong).
Fixed that for you. Jesus christ I don't know where you got that idea from. You can't add clock rates together like that, it makes no sense.
A unidirectional bus running at 3GHz runs at 3GHz
A bidirectional bus with 3GHz signals running in both directions runs at 3GHz
A unidirectional bus with lots of parallel signals running at 3GHz in the same direction is still a 3GHz bus.
The number of wires and directions used do not affect the frequency of the signal.
Your 6750m has a 900MHz 128 bit wide bidirectional GDDR5 memory bus that runs at quad data rate (4 data transfers per clock cycle). This means that there are 3600 MT/s (megatransfers per second) or 3.6 GT/s over the bus with each transfer moving up to 128 bits (16 bytes) of data per transfer. 16 bytes per transfer x 3.6 GT/s = 57,600,000,000 B/s or 57.6 GB/s. It will be listed as 3600 MHz despite the fact that there is no signal anywhere in the bus that cycles faster than 900 MHz because these days manufacturers always list the data transfer rate instead of the clock rate but use Hz as if it were the clock rate (and if anybody points that out to them they just say that it's the "effective clock rate", another BS invented term used for product marketing).
1050MHz x 4 transfers per cycle = 4200 MT/s or 4200MHz "effective clock rate"
Quote:I believe it is x4 not x2
The correct terms are quad pumped and dual/double pumped or QDR (quad data rate) and DDR (double rate). x4 and x2 refer to bus width not transfer rate ratios.
He probably meant his 550 ti which has memory clock of 1025 but sometimes factory overclock to 1050 mhz.
Anything higher then that is "fake clocks" as NV explained
Simply put for gddr5 its real clocks x 4,and sometimes some programs report it as real clocks x2 for some reason...
What I was saying was they probably added it together like you do with dual channel RAM. That was my thought. So I said that the listed clock was 2100, while that was probably just stupid marketing, so the actual clock should be 1050. You told me about this; 1600 MHz RAM is actually only 800 MHz, just dual channel. So, assuming 2100 MHz VRAM was similar, I said 1050 MHz. You completely misunderstood what I was saying. I put the clock rate in marketing terms and in what I thought was technical terms (although apparently I was wrong because VRAM is quad pumped).
I did not no about quad pumped/dual pumped, but I would guess it is quad pumped? I'll check when I get home.
gddr5 is supposed to be "quad channel"
I have no idea why some programs like EVGA precision report "dual channel" speed.
Would like some explanations on that
Its supposed to be either 1050 or 4200 mhz...( your 550ti card )
Quote:What I was saying was they probably added it together like you do with dual channel RAM.
Stop talking, you're only making it worse. This makes even less sense than what you said before. I don't know where you learned to calculate signal frequency but you're doing it wrong. Multichannel interfaces have NOTHING to do with frequency. You do not add together the frequency of each channel. That's like saying a dual core 3GHz processor runs at 6GHz.
Quote:That was my thought. So I said that the listed clock was 2100, while that was probably just stupid marketing, so the actual clock should be 1050. You told me about this; 1600 MHz RAM is actually only 800 MHz, just dual channel.
No it's not. See above.
DDR (double data rate) sdram is double pumped, thus the name. This has nothing to do with channels. Adding more memory modules into more channels does not magically make the frequency go up.
Quote:So, assuming 2100 MHz VRAM was similar, I said 1050 MHz. You completely misunderstood what I was saying. I put the clock rate in marketing terms and in what I thought was technical terms (although apparently I was wrong because VRAM is quad pumped).
No I understood perfectly what you were trying to say. It was just completely wrong regardless of what terminology you tried to use. You clearly have absolutely no idea what you're talking about in regards to this matter. And that's ok as long as you don't pretend to know what you're talking about.
Quote:I did not no about quad pumped/dual pumped, but I would guess it is quad pumped? I'll check when I get home.
I just told you it was quad pumped! Why are you guessing?
Quote:gddr5 is supposed to be "quad channel"
I have no idea why some programs like EVGA precision report "dual channel" speed.
What the hell are you talking about?
1. The bus standards don't define the number of channels. In graphics cards for example we have seen GDDR5 interfaces that are single channel (64 bit), dual channel (128 bit), triple channel (192 bit), quad channel (256 bit), and hexa channel (384 bit)
2. The number of channels doesn't affect the transfer rate in any way. It only affects the bus bitwidth, bandwidth, and throughput
3. EVGA is not reporting dual channel speeds. That makes no sense. The "speed" is usually either the data transfer rate or bandwidth.
I will say this one more time, and if you still don't understand what I am saying then you will never understand it:
1. The way RAM is advertised (not how it actually works) is the sum of all directional frequencies. Does this make sense? Absolutely not. Is that how RAM is quantified by most people? Yes.
2. I assumed VRAM was the same. If EVGA listed 2100 MHz, I assumed they actually meant 1050 because it was probably just marketing.
3. Did I confuse the word "channel" with something else? Yes.
4. I never, ever intended to stipulate that you could add frequencies together to get a true frequency. I said that stupid marketing firms do that, so I listed that since most people would understand that better.
5. I was evidently wrong on quad/dual pumping, as I did not know GDDR5 was quad-pumped or could be.
6. Did I use incorrect terminology to describe memory clock rate? Yes. I used it intentionally because it is what most people are used to.
7. I do know what I am talking about. You just don't know that I'm not saying that it is correct use of the frequencies.
Quote:1. The way RAM is advertised (not how it actually works) is the sum of all directional frequencies. Does this make sense?
No. Because it's not true. Can you find a source for this? Because I have never, ever seen a company do this with any bus, let alone memory buses.
Memory module and other hardware vendors do not adjust reported frequencies based on number of channels or bi-directionality. They adjust it based on data rates (double pumping, quad pumping, etc.).
Quote:Is that how RAM is quantified by most people? Yes.
No. I don't know where you got that crazy idea from because you are literally the first person I have ever run into to come up with it.
Quote:7. I do know what I am talking about. You just don't know that I'm not saying that it is correct use of the frequencies.
No you don't. You clearly don't know the first thing about any of the major memory standards or dram interfaces in general.
I'm looking for the post in which you told me exactly what I am now telling you. It will take a minute.
Basically, I said, "huh, weird, CPU-Z reports my memory clock at 800 MHz" and then you correctly responded "that is the actual clock rate, not the bullshit marketing" or something like that.
I probably did say that but I certainly never told you that it was "800MHz each way" or the listed frequency added the frequency of each direction together.
This is no longer worth my time. It will take too long to remember the words you used. This argument is pointless and way off topic anyway.
Damn it, my CS teacher needs to go get his degree again. He explained dual pumping as sending two signals in different directions, not on different edges of the clock (by which I mean rise/fall).
Half of my point stands, the other half falls. Companies do add together the "effective clock rate" (in this case what I believed to be 2100) while the actual clock rate is lower (what I believed to be 1050). However, the actual clock rate is not two signals going to directions, but two signals traveling along the same clock rise/fall cycle.
Does this sound right to you, then?