High Frequency ram vs LOW timings

X x X

X x X

X x X

Associate
Joined
7 Oct 2018
Posts
93
Now... we can all go by benchmarks right.
But.. my crazy loopy mind been thinking... for a long time.. and AMDMATT done the same thing as my theory.

Ok low cas timings = data is processed sooner due to clock frequencies.. but higher frequencies can in effect cancle that out...

Whats the benefit for going for lower timings vs higher frequency IF it works out the same speed..

High frequency is more susceptible to data loss right?

So whats the downside to low cas timings....

Scientifically and technically?

I think amdmatt had his vega master race system at like ddr 4 1400mhz at some crazy low timings and his system was beating the nvidia ones in wolfenstein 2.......

Does ryzen prefer higher frequency or lower timings?
What does infinity fabric like?
In benchmarks it may work out the same. But matching your ram timings to your infinity fabric timings should be sxxxxxc

Anyone here got infinity fabric O.C results... there was a motherboard that allowed I.F tweeking

There was a good overclocking guide by amd at redgamingtech.... they need to make another one though.. going more technical...
 
Low CAS timing demands ability to react faster to commands from memory.
Increasing clock speed does same by decreasing clock cycle duration.
Hence best memories are usually capable to both tight latencies at more reasonable clocks, or then high clock speed at usual latencies.
And from that perspective it doesn't matter, if it's decrease of latency in clock cycles, or decrease of clock cycle duration, which pushes memory past its limits.
Both result in errors.

For effect to PC's performance there's also lot other factors like memory controller.
Memory controller can even have some internal operating latencies and if certain memory clock is exceeded it switches to slower latencies.
So even if timings of memory itself stay same, it's possible that at some point in increasing memory clock that increase causes slight penalty to total memory access latency.

As far as continuous data transfer rate is concerned, that depends rather exclusively on memory clock speed.
But for real world performance memory access latency is often equally important, at least if transfer rate is at good enough level for CPU.


And in case of Ryzen there's additional complication from InfinityFabric, which is CPU's internal bus connecting CPU cores to memory controller and other parts.
It has its speed in certain fixed relation with memory clock.
Hence Ryzen benefits lot more than Intel CPUs from going those basic 2400 or so MHz memory speeds to 3000-3200 MHz level.
 
EsaT said:
Low CAS timing demands ability to react faster to commands from memory.
Increasing clock speed does same by decreasing clock cycle duration.
Hence best memories are usually capable to both tight latencies at more reasonable clocks, or then high clock speed at usual latencies.
And from that perspective it doesn't matter, if it's decrease of latency in clock cycles, or decrease of clock cycle duration, which pushes memory past its limits.
Both result in errors.

For effect to PC's performance there's also lot other factors like memory controller.
Memory controller can even have some internal operating latencies and if certain memory clock is exceeded it switches to slower latencies.
So even if timings of memory itself stay same, it's possible that at some point in increasing memory clock that increase causes slight penalty to total memory access latency.

As far as continuous data transfer rate is concerned, that depends rather exclusively on memory clock speed.
But for real world performance memory access latency is often equally important, at least if transfer rate is at good enough level for CPU.


And in case of Ryzen there's additional complication from InfinityFabric, which is CPU's internal bus connecting CPU cores to memory controller and other parts.
It has its speed in certain fixed relation with memory clock.
Hence Ryzen benefits lot more than Intel CPUs from going those basic 2400 or so MHz memory speeds to 3000-3200 MHz level.
Click to expand...

So
how do we overclock the memory controller?

Yeah i haven't been hearing a lot about infinity fabric...
Amd locked its timings right?

I.F is perfect for gpu memory interconnect

From what i have seen is it smashes NVLINK into pieces

Increasing clock speed uses more power, heat and makes transistors flip faster right...

Timings make clock cycles execute code faster.....

So lower frequency and low cas latencies should in effect =
 
You must log in or register to reply here.
Back
Top Bottom