Difference between socket 1156 vs 2011.

[Spork Schivago]

Re: Difference between socket 1156 vs 2011.

Okay. I thought he was saying the faster RAM doesn't perform better at all because of the higher latency and was trying to show that it doesn't perform better by comparing FPS in a game. I'm sure system RAM plays a part to some degree, but I'd think the video card would play a much better role in determining FPS. I'd like to see benchmark scores. Maybe with some benchmarking utility specifically designed to test RAM and not CPU or video cards.

[diif]

Re: Difference between socket 1156 vs 2011.

Okay. I thought he was saying the faster RAM doesn't perform better at all because of the higher latency and was trying to show that it doesn't perform better by comparing FPS in a game. I'm sure system RAM plays a part to some degree, but I'd think the video card would play a much better role in determining FPS. I'd like to see benchmark scores. Maybe with some benchmarking utility specifically designed to test RAM and not CPU or video cards.

He was showing RAM speed and latency has very little effect on FPS. His tests used the same CPU and graphics card to remove those from the equation.
There are hundreds of sites showing benchmarks of different graphics cards. You can go for the fastest most expensive or pick the best one for FPS/$.

[ChaosLegionnaire]

Re: Difference between socket 1156 vs 2011.

before u guys jump the gun, u might wanna look at these reviews of performance scaling when overclocking the cpu, ram and both:

X-bit Labs - What is the best memory for Haswell? From DDR3-1333 to DDR3-2933: performance scaling test
Neoseeker - Study of Athlon 64 Overclocking Techniques

the conclusion drawn from this is that, it all depends on how u overclock your system and what programs u run that can take advantage of the added memory bandwidth/speed. from that youtube review u posted and what we learnt from the above reviews is that overclocking the ram without overclocking the cpu provides the least performance benefit. explains why the guy in the youtube review didnt notice a performance increase. u have to overclock the cpu to take advantage of the added memory bandwidth.

[diif]

Re: Difference between socket 1156 vs 2011.

before u guys jump the gun, u might wanna look at these reviews of performance scaling when overclocking the cpu, ram and both:

X-bit Labs - What is the best memory for Haswell? From DDR3-1333 to DDR3-2933: performance scaling test
Neoseeker - Study of Athlon 64 Overclocking Techniques

the conclusion drawn from this is that, it all depends on how u overclock your system and what programs u run that can take advantage of the added memory bandwidth/speed. from that youtube review u posted and what we learnt from the above reviews is that overclocking the ram without overclocking the cpu provides the least performance benefit. explains why the guy in the youtube review didnt notice a performance increase. u have to overclock the cpu to take advantage of the added memory bandwidth.

It seems that article shows that buying the fastest memory will show a minimal increase in certain tasks, 2400 is the sweet spot for speed/$.

If memory is fuel injectors and the cpu the engine, then there is no point in adding bigger injectors until you bore out the engine.

Building a PC is all about balance of parts and it being suitable for it's job.

If you want bragging rights at the pub buy the fastest of everything. The smart guy balances cost/reliability and speed of parts to save £$. His performance in the real world is not much different to the first man but he has money left over for beers and a kebab on the way home.

[Spork Schivago]

Re: Difference between socket 1156 vs 2011.

before u guys jump the gun, u might wanna look at these reviews of performance scaling when overclocking the cpu, ram and both:

X-bit Labs - What is the best memory for Haswell? From DDR3-1333 to DDR3-2933: performance scaling test
Neoseeker - Study of Athlon 64 Overclocking Techniques

the conclusion drawn from this is that, it all depends on how u overclock your system and what programs u run that can take advantage of the added memory bandwidth/speed. from that youtube review u posted and what we learnt from the above reviews is that overclocking the ram without overclocking the cpu provides the least performance benefit. explains why the guy in the youtube review didnt notice a performance increase. u have to overclock the cpu to take advantage of the added memory bandwidth.

Thanks for the link. I don't think the YouTube video was talking about any over clocking. Just non-overclocked RAM. Is the increased speed actually faster. But in the end, it seems he was really talking about does faster RAM increase FPS in games. The title of the video is a bit misleading.


I think he was trying to say, in the video, that there's really no speed difference between something like 1333 MHz and 2400MHz sticks because of the faster RAM needing higher latency and the slower RAM having lower latency.

[ChaosLegionnaire]

Re: Difference between socket 1156 vs 2011.

yes but unfortunately for him, his review isnt so professionally or scientifically done imo. while it may help the layperson in learning that "u dont need high-speed ram. just save your money and get valueram instead", it doesnt really help an enthusiast in understanding exactly why not.

also, his thesis about "faster RAM needing higher latency and the slower RAM having lower latency, therefore becoming a zero sum gain" is not exactly scientifically correct. as the pro reviews have shown, faster ram can overcome the higher timings and it actually results in lower overall memory access latency. however, this lower mem access latency wont be felt or have a noticeable real world impact unless its a huge reduction of several nanosecs or more and the cpu might have to be overclocked to take advantage of this.

if u like learning how to play with ram timings and seeing how much the access latency changes with differing speeds and timings, i recommending u download the maxxmem² memory benchmark utility. it measures single and multithreaded memory bandwidth and access latency.

[Spork Schivago]

Re: Difference between socket 1156 vs 2011.

yes but unfortunately for him, his review isnt so professionally or scientifically done imo. while it may help the layperson in learning that "u dont need high-speed ram. just save your money and get valueram instead", it doesnt really help an enthusiast in understanding exactly why not.

also, his thesis about "faster RAM needing higher latency and the slower RAM having lower latency, therefore becoming a zero sum gain" is not exactly scientifically correct. as the pro reviews have shown, faster ram can overcome the higher timings and it actually results in lower overall memory access latency. however, this lower mem access latency wont be felt or have a noticeable real world impact unless its a huge reduction of several nanosecs or more and the cpu might have to be overclocked to take advantage of this.

if u like learning how to play with ram timings and seeing how much the access latency changes with differing speeds and timings, i recommending u download the maxxmem² memory benchmark utility. it measures single and multithreaded memory bandwidth and access latency.

Wow, thanks! Yeah, I'd love to learn how to play with RAM timings and I'd love to see how much the access latency changes with the differing speeds and timings! I'll be sure to check out that program you mentioned!!! I've generally looked up the optimal settings for RAM timings on the net and just inputted the values they listed in the BIOS. I tried understanding what the different values meant once but it was hard to find information on everything and understand what the definitions actually meant. I mean, the last PC had something like 16 options per stick! Other ones have something like 3 or 4!

There was something about TCycles or something. I remember, lower the number, the better. The options where 2 or 1. Something like how many clock cycles does it take to process a command or something like that. I could run with a setting of 1 if I had 3 or less sticks but if I had four sticks (max number of slots on this motherboard I had), it'd only post with a 2. Asus said that's the way it was designed and if I was going to have all four slots populated, I'd only be able to run it at 2T or whatever.

You don't happen to have any good links that explain all the various options and what they mean, do you? I think some of them are CAS Latency, RAS Latency, etc.

[ChaosLegionnaire]

Re: Difference between socket 1156 vs 2011.

There was something about TCycles or something. I remember, lower the number, the better. The options where 2 or 1. Something like how many clock cycles does it take to process a command or something like that. I could run with a setting of 1 if I had 3 or less sticks but if I had four sticks (max number of slots on this motherboard I had), it'd only post with a 2. Asus said that's the way it was designed and if I was going to have all four slots populated, I'd only be able to run it at 2T or whatever.

i think what u are describing is called the command rate.

You don't happen to have any good links that explain all the various options and what they mean, do you? I think some of them are CAS Latency, RAS Latency, etc.

hmm here u go. long weekend reading for u lol...
Hardware Secrets - Understanding RAM Timings
MSI Forums - DDR3 Memory Timings Explained