Overclock.net banner
1 - 20 of 46 Posts

·
Registered
Joined
·
382 Posts
Discussion Starter · #1 ·
Hi Folks,

I'm about to look into the 'finer points' of overclocking RAM but before I start this deeper and more detailed, research quest, I'd like to get a couple of opinions from some of the more experienced OC.net folks out there.

I have Corsair 1600 2x4 low profile RAM that comes out of the box at 9-9-9-24-2T. Corsair's General specs:

Details for CML8GX3M2A1600C9

Memory Type
DDR3-1600 (Vengeance LP-9-9-9-24*1.5v) Dual Channel
Size
8GB Kit (2 x 4GB) Latency
9-9-9-24-2T
Format
240-pin DIMM Heat Spreader

So the question is; is it worth pursuing an attempt to speed it up? I mean, how much better can I realistically expect it to be considering the other hardware I have?
Thanks
 

·
Registered
Joined
·
929 Posts
In gaming, overclocking your RAM would provide a 1% increase in performance. However, in other applications, overclocking your RAM gives you better benefits. I say: why not? The only thing you lose is a bit of time.
 

·
Registered
Joined
·
285 Posts
Quote:
Originally Posted by wingclip View Post

Hi Folks,
I'm about to look into the 'finer points' of overclocking RAM but before I start this deeper and more detailed, research quest, I'd like to get a couple of opinions from some of the more experienced OC.net folks out there.
I have Corsair 1600 2x4 low profile RAM that comes out of the box at 9-9-9-24-2T. Corsair's General specs:
Details for CML8GX3M2A1600C9
Memory Type
DDR3-1600 (Vengeance LP-9-9-9-24*1.5v) Dual Channel
Size
8GB Kit (2 x 4GB) Latency
9-9-9-24-2T
Format
240-pin DIMM Heat Spreader
So the question is; is it worth pursuing an attempt to speed it up? I mean, how much better can I realistically expect it to be considering the other hardware I have?
Thanks
Its pretty stock ram and you wont get it any better to be worth it.
Quote:
Originally Posted by tahayassen View Post

In gaming, overclocking your RAM would provide a 1% increase in performance. However, in other applications, overclocking your RAM gives you better benefits. I say: why not? The only thing you lose is a bit of time.
What other applications? It wont give him better results in anything apart from benching. If your a bencher, Go for it. But otherwise, there is no REAL WORLD benefit from it.
 

·
Registered
Joined
·
2,101 Posts
It depends on your goals. If you have the time and want to experiment, go for it. Testing has shown that as far as real world application performance gains go, there is little to be had with RAM frequencies above ~1333 MHz. or with tighter latencies as DDR3 RAM is not the system bottleneck that DDR and DDR2 were. Seeing as though your RAM is 2T I would not expect it to OC much or allow tighter latencies but you can certainly try.

http://www.overclock.net/t/1236939/how-low-can-the-corsair-vengeance-1600-go-ram-timings
 

·
Turtle Lives Matter
Joined
·
5,030 Posts
It will help give a few more FPS in CPU-Memory bottleneck areas of games, where you are concerned about minimum FPS.

Here is my Far Cry 2 Benchmark in the busy "Action Scene" using high detail settings @1080p with my FX-4100 @4500MHZ and Crossfire HD 6850's OC'd to 1000/1200:

DDR3-1333 CL8 1T

315

DDR3-2000 CL8 1T

310

What's that a 13% performance increase for minimum?

Memory overclocking is also very beneficial for [email protected]
 

·
Registered
Joined
·
2,101 Posts
The average gain in the data above is only ~9% when going from 1333 MHz. to 2000 MHZ. - with an OC'd Vid card. damric choose to distort the true change in performance by using the Min. framerate instead of the Average framerate. From the above we see a 50% OC of the RAM frequency resulted in +3 FPS average increase which substantiates exactly what the test results have shown over and over again, no significant gain in real world apps. - because DDR3 RAM is not a system bottleneck

The extensive testing conducted by Tom's Hardware, AnandTech and X-Bit's specifically stated "no significant" or "insignificant" gains in Real Apps. If you can't see it or feel it, then it's a moot point. +3 FPS @ 36 FPS is invisible to most people. If it makes you happy, by all means, crank it up to 2000 MHz. Oh wait... most folks can't do that with any AMD CPU other than an FX so they would see even less than 3 FPS average gain, perhaps 1-2 FPS, aka "no significant" gain.
wink.gif


Buying new RAM to gain 1-3 FPS would be a poor investment for most folks even at the current low RAM prices. That money is better spent on a faster CPU/GPU than on new RAM. If you already have faster RAM you can test for yourself and see that no significant gains are possible in real apps. Then you'll better understand.

A small improvement in CPU clock will deliver much better system performance than OC'd RAM above ~1333 MHz. so if you need to run your RAM ~ 1333 MHz. to OC your CPU higher, or run looser RAM latencies to OC your CPU higher - that's usually the best way to go for real world system gains.
thumb.gif
 

·
Turtle Lives Matter
Joined
·
5,030 Posts
AMD4ME,

Had I not overclocked my video cards, or used only 1 video card, or used integrated graphics whatever, the margin gained by memory performance would have only increased, as the CPU would have been doing an ever larger proportion of the work. I can post the results of a single 6850 at stock if you like (actually give me a few minutes and I will post it).

I find it odd that you give out advice on memory when you have no memory in your sig rig nor have never posted any of your own personal memory performance data (maybe you are really an Intel user, or maybe you only own a tablet?).

I invite you to submit a MaxxMem benchmark in the AMD Memory Subforum.
 

·
Registered
Joined
·
2,101 Posts
Quote:
Originally Posted by damric View Post

AMD4ME,
Had I not overclocked my video cards, or used only 1 video card, or used integrated graphics whatever, the margin gained by memory performance would have only increased, as the CPU would have been doing an ever larger proportion of the work. I can post the results of a single 6850 at stock if you like.
I find it odd that you give out advice on memory when you have no memory in your sig rig nor have never posted any of your own personal memory performance data (maybe you are really an Intel user, or maybe you only own a tablet?).
I invite you to submit a MaxxMem benchmark in the AMD Memory Subforum.
Making the discussion personal does NOT change the objective test data results - as you have been advised in the past.

Your own data confirms what Tom's Hardware, Anandtech and X-Bit Labs extensive testing confirmed - "no significant system gains" with RAM frequencies above 1333 MHz.

Thank you for confirming what I have been telling you and others for several months. If people actually read the test at Tom's, AnandTech's and X-Bits or conducted their own test, they will get the same results - "no significant gains in real world apps".
thumb.gif


Have a nice day.
 

·
Turtle Lives Matter
Joined
·
5,030 Posts
Quote:
Originally Posted by AMD4ME View Post

Quote:
Originally Posted by damric View Post

AMD4ME,
Had I not overclocked my video cards, or used only 1 video card, or used integrated graphics whatever, the margin gained by memory performance would have only increased, as the CPU would have been doing an ever larger proportion of the work. I can post the results of a single 6850 at stock if you like.
I find it odd that you give out advice on memory when you have no memory in your sig rig nor have never posted any of your own personal memory performance data (maybe you are really an Intel user, or maybe you only own a tablet?).
I invite you to submit a MaxxMem benchmark in the AMD Memory Subforum.
Making the discussion personal does NOT change the objective test data results - as you have been advised in the past.

Your own data confirms what Tom's Hardware, Anandtech and X-Bit Labs extensive testing confirmed - "no significant system gains" with RAM frequencies above 1333 MHz.

Thank you for confirming what I have been telling you and others for several months. If people actually read the test at Tom's, AnandTech's and X-Bits or conducted their own test, they will get the same results - "no significant gains in real world apps".
thumb.gif


Have a nice day.
Not sure what you are talking about as far as advising, but I sure wouldn't be advising anyone on anything if I didn't have my own first hand results to back it up. No one here cares that you parakeet random crap from other tech sites.

Again I cordially invite you put your money where your moth is and submit your MaxxMem result to the AMD memory subforum.

Have a nice day in the "real world" (wasn't that an MTV show?)
 

·
Registered
Joined
·
2,101 Posts
Quote:
Originally Posted by damric View Post

Not sure what you are talking about as far as advising, but I sure wouldn't be advising anyone on anything if I didn't have my own first hand results to back it up. No one here cares that you parakeet random crap from other tech sites.
Again I cordially invite you put your money where your moth is and submit your MaxxMem result to the AMD memory subforum.
Have a nice day in the "real world" (wasn't that an MTV show?)
In previous threads you tried to claim the results from 1200 MHz. and 1800 MHz. RAM frequencies showed substantial gains. I advised you that the discussion was specifically about RAM frequency at ~1333 MHz. and higher.

We see from both the references I mentioned and your own data that +3 FPS @ 36 FPS is "insignificant" - and you need to be able to OC the RAM 50% to 2000 MHz. to get this tiny increase. The test results speak for themselves.

Posting my test data would just lead to another pissfest. That's why I referenced Tom's Hardware, AnandTech and X-Bit Labs as they all did extensive testing on DDR3 RAM frequency and latencies and reported "no significant" system gains in REAL APPLICATIONs above ~1333 MHZ. These are reputable PC industry sources so you should not go around disparaging them when your own test data substantiates their results of "no significant gains in real world apps"

Insults don't change the test results so I'd suggest you stop the childish behavior. Your own test data shows no substantial gains.

Having a pissfest over your test data confirming what I have been saying for months is not going to change reality. It is what it is.

Next you'll want to argue over the definition of "substantial". Instead I suggest you and everyone else run your RAM at whatever frequency makes you happy. The test data is quite clear so you can draw whatever conclusions you desire from it.

Going thru the same debate in every thread does not change reality.
redface.gif
Tom's Hardware, Anandtech and X-Bit Labs all expected to see tangible gains in real world applications with higher RAM frequencies than ~1333 MHz. but it simply does not happen on AMD or Intel desktop PCs. AMD Llano APUs are the exception because the GPU gets to take advantage of the increased frequency.
 

·
Premium Member
Joined
·
20,625 Posts
A +3FPS gain from 36FPS is an approximately 9-11% (Corrected: referenced the wrong calculator app on my system, I find in my math needs I keep several open at once) gain. By no means is that insignificant, especially in certain games like FPS games. Talk to gamers; I'm sure a lot of them will agree. Some of them won't; but for many others, this is an important difference.

There are real applications that are memory-bandwidth sensitive; others are not memory-bandwidth sensitive. But I think that generalizing that all real applications see no gains is a bit short-sighted with all due respect. That has to do with how that particular real-world application interfaces with memory, how it is coded to favour speeds of memory, and how it is coded to favour speeds of the CPU or core count as well as in the use of throughput between the CPU and RAM and elsewhere.

Consider WinRAR: as an uncompression program, all operations which include the constant random transfer of uncompressed and compressed RAM to and from the CPU at the same time (transfer that is larger with a CPU overclock, where more throughput is allowed); it is fairly memory intensive and the gains through the increase in RAM frequencies are hardly insignificant.

This is from an Anandtech review (Note that this is on a Core i7 9XX LGA1366 platform and is not easily comparable to results on AMD systems):

winrara.jpg

There is a 10% time jump from 1333Mhz and 1600Mhz at CL9 latencies and down from the low end 1066 @ CL7 to 1866 @ CL7 there is a 23% difference.

Both timings and speed/bandwidth increase generally make a difference (although timings show very little difference as is observed by most Intel platforms; on AMD platforms timings make a much bigger difference (i.e. actually modifying bandwidth in as significant a way as speed) in general memory performance).

It should be noted that these tests make use of 6GB memory kits running in triple channel config. By the time you reach triple channel config, the amount of throughput is so redundantly high that memory usually does not see significant gains; but that is not true in these results, where differences still appear to be recorded despite the availability of much redundant throughput and bandwidth. These lines should be much more amplified (i.e. the highs higher, the lows lower) on dual channel platforms.

This might not be considerably large at first. Consider however that this tests the uncompression of a single ~900MB file. Where users may be extracting several RAR files at once (i.e. several gigabytes' worth in split RARs) or larger RAR files - at which point, the time savings can become very significant - in the matter of several minutes to even several hours if involving terabytes' worth of compressed data (i.e. hard drive backups).

This is H.A.W.X.:
hawxna.jpg

The gap between 1333 CL9 and 1600 CL9 is $0-5 and a 10% minimum FPS gain in this game on this platform. This is hardly insignificant. For gamers this is respectable gain that may be highly valued.
 

·
Registered
Joined
·
382 Posts
Discussion Starter · #13 ·
First I would like to thank you all for your replies on this question and though I really didn't want it, (the question), to spark any tempers the information I got from both sides is invaluable for any future decisions I may make. I know I'll be referring to these posts as I proceed to learn about these points.

Second, I almost exclusively fly FSX, (at present Intel seems to have the best CPU's for that game), and I use Dragon Naturally Speaking Premium 11 a lot as well. I notice that the latter tends to hit my memory pretty hard when I have several programs running, (internet research as well as background programs).

FSX is a very CPU intensive game but it also seems to hit the memory pretty hard. I recently installed the ASUS mobo and the Corsair RAM. My old mobo was a HP Stock board and the old RAM was 4x2 Seagate 1333 and I forget the exact numbers but the latency was double digits all the way down, (something like 13-13-13-38).

After the installations I can say that the performance of FSX was incredibly better even though my CPU was not O/C'd any more than it is now, (about 4.1GHz on both boards). Dragon also showed less 'pull' on the memory gages. This can all be attributed to the mobo and memory combined for sure but I can't say which played the greater part because they were installed simultaneously.

As for fps, I have my FSX game locked to 34 fps in game and I can say that it was maintained much more often, (went from an average 19fps to 28-30fps in the more 'demanding' areas) with the new installs.

Thanks for all the help and I will take all in consideration.
thumb.gif


PS I stopped receiving notifications when someone answers my posts recently. I have everything set to on but my stupid AOL email must have started blocking these like it has many of my other forum websites. So I can only check back every so often to see if anyone posted. So if I don't reply quickly to a post, please understand that this would be the reason why. Thank You
 

·
Registered
Joined
·
285 Posts
Quote:
Originally Posted by xd_1771 View Post

This is H.A.W.X.:
550x492px-LL-8e348bb2_hawxna.jpeg
The gap between 1333 CL9 and 1600 CL9 is $0-5 and a 10% minimum FPS gain in this game on this platform. This is hardly insignificant. For gamers this is respectable gain that may be highly valued.
Seriously?
biggrin.gif
HAWX
biggrin.gif


I've actually been playing that game quite abit later, and get this, On my 955BE with mismatched memory (Corsair budget and G.skil somethingorather)running CL9 at 1333mhz and a GTS 250, i get 59-60fps with everything set on high.....Then! with my FX 6100 on stock, My 2gb Dirt 3 6950 on stock and 8gb of Dominator ram, Running 1600mhz on CL9, i still get 59-60fps max and min with everything set on High.

No difference, even with the graphics cards
biggrin.gif
and the increase you mentioned in that game, a whole 3fps, yes, it is very insignificant. If it was 10plus, then i'd make a big deal, but an extra 3fps that your not going to notice, its not even worth of note.
 
  • Rep+
Reactions: AMD4ME

·
Registered
Joined
·
2,101 Posts
wingclip-

Unfortunately we get the same meritless arguing in many RAM threads because folks are unable to deal with the reality that their beliefs are unsupported by objective test results from reputable PC industry sources which are the same as the results of numerous OCN members who have also conducted real world app. testing. Some folks do not understand that synthetic benches are not the same as real applications. That's why this discussion is specifically in regards to real apps where no significant gain is observed.

As an example of the problem... 3.29 FPS gain @ 36.22 FPS video speed is not a 12% gain in FPS, it's a 9% gain... This should illustrate the problem. (Divide 3.29 by 36.22 and see what you get for a result).

Independent testing has shown that above 30 FPS most people have no idea what the true FPS is so it doesn't matter as the eye can't really tell. The discussion is about "substantial" gains or not. If you can't see it or feel it, it doesn't really matter. The test data from Tom's, AnandTech, X-Bit Labs and "damric" all show NO substantial gains in real world applications.

I highly recommend that people conduct their own testing at 1333 MHz. and then at whatever the highest RAM frequency they can run stable and see for themselves there are no substantial gains in real world apps.

This is NOT a conspiracy amongst the sources who all independently reached the same conclusion - it's just reality. If you understand that DDR3 RAM @ ~1333 MHz. is NOT a system bottleneck and that one clock cycle in REAL TIME becomes less as the frequency increases, then you'll understand why there is no significant system performance gains to be had in real apps. "Damric's" 50% RAM OC to 2000 MHz. resulted in 3 PFS @ 36 FPS video - which is nothing. If you don't understand the technical aspects of the discussion then you'll just continue to argue and post misleading information in spite of the overwhelming objective test data that shows no significant gains in real apps. It is what it is folks. Next people will want to argue over the definition of "significant"... Most folks would agree that if you can't see it or feel it, it ain't significant. Blind testing would prove this to be the case but it still would not change some people's opinion.

Arguing and insults don't change the test results. My goal is to present accurate technical information. I am fully aware some folks take exception to this information yet no one has shown any proof that the test data referenced in this case by three independent PC industry sources is incorrect. In fact "damric's'" own Far Cry 2 test data substantiates exactly what I have been saying and what other OCN members have confirmed from their own testing along with Tom's, AnandTech and X-Bit Labs. Do people here REALLY believe these three PC industry sources are conspiring to mislead enthusiasts? REALLY ?

BTW, you'll note that no where have I told people to NOT OC their RAM. What I have done is advise them of the results they can expect based on extensive objective testing by the sources listed above. This way when they see no significant gains in real applications - they will understand WHY this is the case.

Good luck with you system wingclip.
thumb.gif
 

·
there is no easy way out
Joined
·
3,284 Posts
As currently I have no real world comparison, but at least I do have some synthetic benchmarks of different frequencies, which proves significant increase in numbers.

1333 9-9-9-24-2T, IMC/CPU-NB = 2000Mhz, CPU 2.8Ghz (Default clocks)
452

1333 6-7-6-19-1T, IMC/CPU-NB = 2500Mhz, CPU 3.25Ghz
447

1526 7-8-7-22-1T, IMC/CPU-NB = 2860Mhz, CPU 3.28GHz
453

1652 7-9-7-23-1T, IMC/CPU-NB = 3100Mhz, CPU 3.25Ghz
424

I don't care whatever the result would be...but I'll surely post with some real world applications very soon.
 
  • Rep+
Reactions: damric

·
Turtle Lives Matter
Joined
·
5,030 Posts
AMD4ME, XD, Wing, Sorry I wanted to get this post to you guys sooner, but I had taken the kids to the beach today, and it must be Spring Break here on the Gulf, because all the young hotties were out. I'm glad I wore my sun glasses, even though they didn't prevent my wife from smacking me as I stared at the perfectly shaped floatational devices.

Anyways, I re-ran some tests, and ran a couple 3dMark11 combine tests, because AMD4ME was right, and one of my posts above was actually at 1200MHZCL8, and my GPUs were indeed OC'd heavily. Now what I'm giving you is some samples with and without my GPUs OC'd, and memory at 1333CL9 and memory at 2000CL8. CPU at 4500MHZ, and driver and detail settings identical for all tests, with Crossfire HD6850's.

First let's look at some more FC2 "Action Scene" (Yeah this game is old, but it comes with a great benchmark tool!!!) I'm including my older benches of 1200, 1800, and 2000 again as well as the new ones I ran last night. The "Action Scene" is exactly that, very intensive, and generally forces a CPU bottleneck. I set each benchmark to run 5 times, so we are looking at the average score.

DDR3-1200CL8 GPUs at 1000core/1200mem

315

DDR3-1333CL9 with GPUs at stock

338

DDR3-1800CL8 GPUs at 1000/1200

309

DDR3-2000CL8 with GPUs at stock

418

DDR3-2000CL8 with GPUs 1000/1200

310

Next we move on to some 3dMark11 "Combined Test" @performance level preset. All of these tests were run today, and I only ran each setting once.

DDR3-1333CL9 GPUs Stock

338

DDR3-1333CL9 GPUs 1000/1200

338

DDR3-2000CL8 GPUs Stock

338

DDR3-2000CL8 GPUs 1000/1200

338

Ok, now go back and look again carefully this time.

Overclocking the memory gave a larger performance boost than overclocking the GPUs!

This is clearly because the CPU was bandwidth starved during these CPU-intensive benchmarks, causing a bottleneck on the GPUs!
 
  • Rep+
Reactions: sumitlian

·
Registered
Joined
·
2,101 Posts
Thanks for the Far Cry 2 data "damric". It confirms exactly what Toms Hardware, AnandTech and X-Bit Labs has stated - there are no substantial gain in real world apps with increased frequencies or tigher latencies above ~1333 MHz. because DDR3 RAM @ ~1333 MHz. is not a system bottleneck for typical desktop PCs.
thumb.gif


Llano can use the increased frequency for the GPU however.
 

·
Premium Member
Joined
·
20,625 Posts
Generalizing that every gain creates "insubstantial" or "substantial" results in commonly-used real world apps is preposterous. One must consider the large variety of real world applications that suffer from different conditions. RAM is not like CPU or GPU where data is simply processed and a level of performance can be easily measured; in RAM, operations are more complicated; data must be identified as processed or unprocessed, created, deleted, read to, written from, and moved constantly in random orders. In many cases, it is simply completely inappropriate to compare results with benchmarks; noticeable differences that are gained through the increase in RAM frequency and the general increase in bandwidth at times cannot be measured.

But if we were to be taking it easy and talking about measurable differences, then again, I would like to refer to my WinRAR example. Those gains create huge differences when working with exceptionally large files.

Evidently, not every program benefits "significantly" through faster RAM. But that does not mean that every program will not benefit from it - including apps that are commonly used in real-world scenarios. There is nothing misleading about the information concerning gains that do exist that we have provided. Your generalizations and perception of what is "substantial" is something that is generally personal and that may be misleading in that not everyone can agree with you.

-----------

Concerning the game difference that was pointed out by damric, the 3FPS difference of approximately 9% recorded was for the average FPS. When we look at the minimum FPS, that is a 4FPS and nearly 14% gain; when considering it is minimum FPS at play, this can be hardly, hardly insignificant. These FPS differences have more to do than in just gameplay smoothness; it can also define a lot of other things such as amount of tearing. If this were the difference between 52 and 60 FPS (14%) minimum.... this could be the difference between a slightly jerky game with tearing to a totally smooth and fluid game on a 60hz monitor.

The fact that such FPS gains in certain games, however, can be recorded from a RAM improvement alone should be recognized as exceptional feat considering that if DDR3 at 1333Mhz were - as it is so claimed - "not a bottleneck", there wouldn't be any gain whatsoever, as the GPU & CPU would in normal cases foot most of the bottleneck. Any visible, constant performance hold-back upon heavier loads is a bottleneck; in this case, it may not be the same kind of bottleneck that is seen when CPUs & GPUs bottleneck programs, and yet is there and it is creating what could be considered significant difference depending on the gamer and the type of game. Far Cry 2 is a first-person shooter; in these games, players need to act fast and are very sensitive to any FPS and latency changes.

-----------

Generally one must have a developed knowledge of how programs work to be able to forecast whether any significant difference could be required through RAM speeds. For example, wingclip noticed a significant FSX difference with faster RAM. Attempting to stir his direction elsewhere by taking our arguments and calling them fallacious will do nothing because he already has confirmed that there are benefits to faster RAM in what he does, which jutsifies additional OC. This can be supported through an examination of how FSX works.

In FSX, this may have to do with not only the requirement of having to render large amount of different small objects (which probably represent ground objects such as trees, buildings - some of which move) and having to know the exact location of every one of these objects on the virtual map but also having to load their specific texture files from the HDD and ensuring they are applied to the correct sprite - this can be exceptionally bandwidth-intensive. Where very large chunks of data might need to be transferred between RAM cells, speed of RAM can define whether the game experience is smooth or whether there is a lag in delaying textures and polygons (which could also delay gameplay to some extent). The "demanding areas" he describe may be areas that have more polygons in view of the camera in addition to several overlay texture with alpha/transparency maps including clouds.

I know and can predict this having limited experience in Microsoft Flight Simulator and knowing how it works, and the same is true in 3D view city-building games such as Cities XL by Focus Home Interactive.
 

·
Registered
Joined
·
2,101 Posts
^^^

I knew when "damric's" Far Cry 2 test data showed no "SIGNIFICANT" gains, you'd be along to argue the definition of "significant". If you believe 3 FPS @ 36 FPS is "significant" - great. Most people would disagree with your conclusion if you can't see or feel the change. 10,000 words in this thread isn't going to change reality any more than it did in the last RAM thread that was locked when you refused to read the test data from Tom's Hardware, AnandTech and X-Bit Labs, yet continued to post questions that were answered in the referenced test data.

Damric's test data shows that a 50% OC of the RAM to 2000 MHz. resulted in 3 FPS @ 36 FPS. It is what it is - insignificant as it can't be seen or felt by most people. As "madengineer" previously posted, if there was a 10 FPS increase at 36 FPS that would be considered a "significant" gain, but 3 FPS with a 50% increase in RAM frequency is a joke really.

"windclip" made both a mobo and RAM latency change from a very unusual CL 13 setting that normal RAM does not use. He admits he has no idea what caused the changes - which were undocumented so clinging to a "wing and a prayer"... doesn't change anything.
wink.gif


For some inexplicable reason you refuse to accept the test data from Tom's Hardware, AnandTech and X-Bit Labs and instead try to use synthetic test benches to support your beliefs even when it has been mentioned over and over that we are discussing REAL APPLICATIONS not synthetic benches. No one is disputing small gains in SYNTHETIC benches - which are theoretical. Using REAL APPS confirms there are no "significant" gains.

I have suggested that all OCN members run Real Applications at ~1333 MHz. and at the highest their RAM will run stable and see for themselves what the gains are. Then they can decide if something they can't see or feel is "substantial" or not. It would be great fun to do blind tests in person because some people are hard to convince.

Since I know you're not going to change you POV and the test data is not going to magically change to show a substantial gain in REAL WORLD APPLICATIONS, I suggest we Agree to Disagree since everyone here is fully capable of running their own test. Arguing your beliefs is not going to change the test results. It is what it is.

I see you corrected your math error above... That's a start to accepting the test data for what it is.
thumb.gif
 
1 - 20 of 46 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top