Since the tech gurus are here...

[pmr]

I reposted in the cpu forum.

I remember two years ago when I had a Pentium 4 running at 3.6Ghz and ram running at 400ish.
Technology appears to evolve every month and some of us right now have VRAM running faster than CPUs and maybe in a few months even the normal RAM.

Modern CPUs can process more work in less cycles, contrary to memory that keeps going up and up.
So, is there any new technology in sight to revert this Mhz freaks(if even possible) and keeping bandwidth high with lower freqs.? Or the only way to bandwidth is pure Mhz?

This question may be stupid, but, whatever... :?

 

[pmr]

anyone?

 

[1Tanker]

anyone?

I don't think it's a stupid question. I don't know the answer, but bus-speed, cache size and speed should all contribute to RAMS speed(less bottlenecking). Also, getting latencies as slow as possible should help speed up memory access(as L1/L2 do). But i don't know of any "architectural changes" that can increase RAM's speed other than.....raw speed in MHz. DDR was a breakthrough in RAM throughput, so if they can quad-pump it, or something, i guess that would help. :?

 

[m25]

Well, actually we do need some revolution regarding RAM, but that would be more price wise than anything else. Currently, DDR2 offers enough bandwidth even for the best CPUs and cache RAM is far more important. However, we all saw the CPU price drop and if RAM prices could drop like that, it would be thr best news of all, beyond performance and tech advancements.

 

[AdamBomb42]

Something certainly has to be done about the memory issue. My bother who is working on the Thurley platform has informed me that the DDR3 (effective 1333Mhz) that it is using has a lot of current running thru it and that they get very hot. We are quickly coming to a point that some real innovation needs to happen with memory.

 

[m25]

The most logical derivation seems some QDR RAM, it would double the bandwidth without much effort and power consumption; the same trick that DDR did over SDR. I don't understand why they pursue e 1.3GHz DDR3 when a 800MHz QDR would perform even better.

 

[cmptrdude79]

Either that or widen the bus. The two factors in speed here are the bus width (how much data you can transfer per time) and of course the actual frequency at which it is operating.

Since QDR would probably take quite some time to get right, I would think the best short-term alternative would be a wider bus. There are all sorts of other issues that would have to be addressed, but that's where I think the advancements will come.

-J

 

[bluntside]

The last couple of Issues of CPU aka Computer Power User, Freescale have been working on Electro magnetic flash ram using high power Neodymium magnets, and applying them into ram< I have no Idea how this works, but with this ram there will be no need to wait for your pc to "boot up" Once you turn it on it will all be there as if you opened you gridge and you find the light already turned on.

 

[sweetpants]

I reposted in the cpu forum.

I remember two years ago when I had a Pentium 4 running at 3.6Ghz and ram running at 400ish.
Technology appears to evolve every month and some of us right now have VRAM running faster than CPUs and maybe in a few months even the normal RAM.

Modern CPUs can process more work in less cycles, contrary to memory that keeps going up and up.
So, is there any new technology in sight to revert this Mhz freaks(if even possible) and keeping bandwidth high with lower freqs.? Or the only way to bandwidth is pure Mhz?

This question may be stupid, but, whatever... :?

I actually heard that currently they are working on carbon nantube flash memory modules... Something that could be revolutionary I suppose .

 

[m25]

I actually heard that currently they are working on carbon nantube flash memory modules... Something that could be revolutionary I suppose .

We hear such things since 4-5 years ago, also regarding flash HDDs but AMD's fusion and Intel's 32nm will come sooner than any significant change in memory happens .

 

[sweetpants]

I actually heard that currently they are working on carbon nantube flash memory modules... Something that could be revolutionary I suppose .

We hear such things since 4-5 years ago, also regarding flash HDDs but AMD's fusion and Intel's 32nm will come sooner than any significant change in memory happens .

Oh most definately, the only reason I brought it up because there was a recent article on it but I can't remember where. I believe a university was making some progress on it.

 

[SSS_DDK]

there are a lot of memories out there, some revolutionary, some more classic. Magnetic Ram, Biological memory, Nanotubes, etc... however whether they will find their way into computers!???!!!!!!Somebody knows but it ain't me.

 

[sweetpants]

Ahh you jogged my memory... I was mistaken

The memory they were working on was NVRAM rather than traditional volatile ram that is in your system.

 

[Pippero]

The main point of hight MHz is to improve on latency (reduction of the cycle time), not bandwidth.
Of course, they can also be used to increase bandwidth, because if you transfer the same amount of data per clock, and you increase the rate of transfer, then you increase the amount of data transferred in the time unit (Doh! )
The other way to increase bandwidth (but which does *not* reduce latency) is to increase the memory interface and memory bus width.
While conceptually this is easy, practically there are several problems, as with routing all the wires and especially you have problems with interference and clock skew.
Clock skew means that data traveling on parallel lines on a bus, which starts all at the same time, arrive at slightly different times at destination due to differences in length of the lines or line impedance or interference.

 

[sweetpants]

Clock skew means that data traveling on parallel lines on a bus, which starts all at the same time, arrive at slightly different times at destination due to differences in length of the lines or line impedance or interference.

Difference in length shouldn't be happening as all traces on a motherboard (and most likely memory) are the same length to avoid that problem. You're most likely looking at impedence as the main culprit, and interference.

 

[aBg_rOnGak]

I seems to agree with your opinion,especially because years ago,I was thinking the same thing.If it's currently 8-bit wide,why not increase to 16-bit,32-bit or even 64-bit?QDR also sounds good provided the technology is there and has matured enough it can be released to consumers.But mind all of you,I'm not well versed in memory technology,all that I wrote was purely opinion and hopes

 

[mr_fnord]

I reposted in the cpu forum.

I remember two years ago when I had a Pentium 4 running at 3.6Ghz and ram running at 400ish.
Technology appears to evolve every month and some of us right now have VRAM running faster than CPUs and maybe in a few months even the normal RAM.

Modern CPUs can process more work in less cycles, contrary to memory that keeps going up and up.
So, is there any new technology in sight to revert this Mhz freaks(if even possible) and keeping bandwidth high with lower freqs.? Or the only way to bandwidth is pure Mhz?

This question may be stupid, but, whatever... :?

What RAM is running at Ghz? There are doublepumped busses and such, but the RAM is running from 400-900 Mhz, and it is very expensive to boot.

You can keep Mhz low and increase memory bandwidth, hence 128 bit, 256 bit, dual channel, etc. memory. Double data rate was another innovation, allowing double the bandwidth on the same pipe. However, these technologies add latency, or at least they don't improve it. Looking at DDR vs DDR2 RAM, you'll see DDR2-1000 memory (500 Mhz bus), with latencies of 5-5-5-15, while DDR-400 (200Mhz bus) had latencies of 2 or 2.5. This is roughly the same number of nanoseconds, so the individual memory cells did not double in performance, but the external communication rates increased. Internal cell latency is steadily improving with improved materials science, and there are all of these breakthrough technologies that are supposed to dramatically improve memory performance, like ZRAM or RDRAM in days of yore, but often by the time they reach market they are only a 10-20% improvement over the existing designs. That was one of the problems with RDRAM, it released designs that were groundbreaking, but by the time it was licensed and in mass production, DRAM had caught up to it then surpassed it.

IBM and Intel (and a ton of others) have shown variations of embedded memory running at CPU clock speeds, solving latency problems and dramatically improving memory access times, but memory is a fairly large silicon structure, so making memory at the same process level as CPUs is rather expensive. Articles about IBM's eDRAM indicate that 128MB is roughly the same size as a CPU, and a commodity CPU optimistically costs ~$50, so that's $.40/MB. Compare to DDR2 at $.08/MB and you'd be hard pressed to find someone who'd pay more for the memory subsystem than the rest of the system. The exception is in GPU's, where VRAM often costs as much as the GPU itself, but that low latency is needed for performance.

Of course, increasing internal memory speeds to CPU performance doesn't help on a very slow bus. If total memory latencies were 1 bus clock, which is impossible, there would still be a 14-24 clock (based on CPU multiplier) delay between the end of the memory request and the start of the transfer. Better than 150+ today, but still not perfect. IBM and Intel have discussed and demonstrated designs with embedded memory so that memory speeds are close to cache speeds, but then you'd be buying a processor with a set amount of memory in it, paying a ton of money for that memory, and upgrading memory would mean getting a new processor.

 

[thexder1]

there are 4 things that affect how fast data move between devices. this is the same for all devices.

1. frequency the data travels at (MHz, GHz)
2. width of the connection (I believe DDR is 128 bits)
3. latency (the amount of time it takes for the data to travel or in the case of RAM the time it takes for it to settle before you can do something with that area of RAM again)
4. overhead (this mainly applies to networking but there is overhead in all the communications)

If the memory doubles in the width of the connection you will get roughly double speed (hence video cards using 256 and 512 bit memory busses) if you double the frequency you will double the speed. the latency and overhead are not as simple calculations as the other two but the lower latency and lower overhead the higher speed. I never really understood why there was a big push to DDR2 when DDR 400 is pretty much the same speed as DDR2 800. DDR has been up as high as about 650 MHz which would be roughly the same speed as DDR2 1300 because of the high latency of DDR2. The only really good part of DDR2 is that it is lower power than DDR and they claim that it is cheaper to make but the prices do not seem to reflect that.

 

[pmr]

What RAM is running at Ghz?

DDR3 and DDR4 VRAM in gfx cards

 

[Pippero]

DDR2-800 has the same latency as DDR-400, but theoretically double the bandwidth.