Should we consider RAM content?

[Rob Williams]

When DDR and DDR2 were at their prime, we posted reviews around memory kits often. The real meat of those articles focused on overclocking, because back then, faster memory could actually be utilized to some extent. Then with the advent of DDR3, frequencies have hit unbelievable heights and latencies are very reasonable, so it begged the question... what's the point? We've reviewed a few DDR3 kits, but after those few, it hit me... and I didn't see a strong enough reason why memory kits should be reviewed at all, so we stopped publishing such content.

Essentially, what it came down to was that I felt if we reviewed memory, it would be directly aimed at the overclockers, because they're the ones who obviously care the most about faster parts. But that's a limited market, obviously, and because of that, attention was better spent elsewhere.

But, with the launch of Core i7, my mind began to be swayed just a little bit, and I'm wondering if it might be worth getting back into doing some memory-related reviews again, but I'd like some input. The reason I like Core i7 so much for memory is that you can adjust the clocks of the kit without touching the CPU clock speed. That was a huge issue with previous platforms, because as the RAM became faster, so did the CPU speed. So, how on earth is someone to realize the actual benefits of faster memory?

Dual-Channel kits are still a problem though, because previous platforms are locked in with the same limitations I mentioned above, both AMD and Intel. You can use a Dual-Channel configuration on a Core i7 machine, but the results wouldn't be near as typical as what people would see on a real Dual-Channel motherboard, so it's pointless to benchmark them that way.

So I guess with this idea, we'd be locked into Triple-Channel DDR3, which is too bad, because that excludes the vast majority of the market. But like I mentioned, it's far too difficult to reliably gage the worth of a Dual-Channel DDR2 kit when something other than the memory frequency has to be altered in order to achieve the speeds we need... either the CPU clock or the FSB.

But all of that aside, are triple-channel memory reviews something you guys would be interested in? I receive press releases all the time showing off new kits, and for the most part, they're all the same, however... it might be a good idea to pit a bunch together in a roundup and see who comes out on top from various standpoints (ultimate performance vs. ultimate price). Heatspreaders could also be taken into consideration... their design and how well they cool.

Of course, I will not do a roundup or any memory review if I can't figure out a good test suite. When I posted our "Choosing the Best Memory Kit" for the Core i7 platform, the benchmarks didn't really show much of an improvement at all between frequencies. So, that just means we need to figure out what applications DO benefit from higher memory frequencies. Adobe Lightroom was one fo the few here to see any variations, but I'm sure there's more.

Thoughts?

 

[Kougar]

Memory reviews still have some usefulness... almost any average enthusiast will want to know how much headroom various memory kits offer. I'm not talking about 2,000Mhz kits, but average 1333Mhz and 1600MHz kits nobody wants to push beyond 1.65v Especially 1333MHz, since some can hit 1600 without any trouble or additional voltage.

My brand loyalty to Corsair has ended, so I was careful before taking the plunge with a kit of OCZ Platinum. It's cheaper, overclocks better, and offers lower timings than anything Corsair offers. After noting Corsair had redesigned the heatspreader on DDR3 Denominator memory making it "cheaper" I lost my long standing desire for it. The cheapest Denominator 1600MHz kit is twice the price of the OCZ Platinum 1600MHz kit, and still has higher timings to boot.

I didn't read the rest of the review but this single page is exactly what I look for in a memory review... emphasis goes for "roundups": Link I don't see any point about reviews for memory performance... any kit is as same as the other if the specifications match.

 

[Rob Williams]

It's true that an enthusiast will prefer faster memory, but I've yet to see a real purpose for that. That's the issue. Faster CPUs offer obvious benefits, but what does a DDR3-1800 kit offer over a DDR3-1333 one, seriously? What does it improve? Where can the normal person see an improvement thanks to the faster kit?

If anywhere, workstation loads might be able to take advantage of the extra speeds, so that's likely what I'd focus on... none of this synthetic benchmark nonsense... we need REAL and beneficial benchmarks.

I don't see any point about reviews for memory performance

That's the point. What's the purpose of a review that focuses just on overclocking? To increase our reader's e-peens? There has got to be a way that faster memory proves beneficial, else these fast kits are all for naught.

Interesting about the Corsair vs. OCZ thing... I didn't realize that.

 

[Kougar]

Rob, you are kind of missing my point! I agree, there's no real purpose for buying 1800MHz, 2000Mhz, or 2133MHz memory. So if the average Joe is smart he will buy a 1333MHz or 1600MHz kit (because 1600MHz RAM is actually cheaper). The question then becomes for average Joe, what 1600MHz kit holds the best value, can offer the tightest timings below official specs, and most overclocking headroom should they wish to overclock their Core i7? I'm not talking performance numbers, just results of identical 1333MHz and 1600MHz kits pitted against each other for headroom and tightest timings only. Wouldn't even need benchmarks, just some programs to stress test and stability check the RAM settings.

Say for example if I'm going to buy a 1600MHz kit... I'd want to buy the best one if they all cost the same, right? That is what I am trying to get at here. There are a dozen 1600MHz kits with less than $20 difference between them, and people wouldn't expect the cheapest kit (OCZ Platinum) to also be the absolute best kit with the tightest timings offered. 1600MHz at 1T 7-7-7-21 timings is perfect... not even Corsair's Dominator kits can do that.

Edit: If I spell DOMINATOR as denominator one more bloody time...

 

[Rob Williams]

I didn't miss the point, it's just that I still feel there is so little point in having faster RAM. I agree though, timings are what matter most, so that would definitely be a major topic in such a roundup. Thanks for the input, I really appreciate it. Our goal is obviously to be a little different, RAM reviews is where that becomes difficult. It's HARD to prove to people that faster memory is worth the money, and in truth, it always has been.

If I spell DOMINATOR as denominator one more bloody time...

Haha! Don't worry, I do things like that often.

 

[Kougar]

If I plan to overclock my Core i7 920 beyond 4,000MHz, then my RAM would be forced to run beyond 1600MHz... therefore, making sure my RAM can do so at 1.65v was an important consideration. Knowing my OCZ Platinum should reach 1800MHz at CAS 8 means I will have all the headroom I need for overclocking the new D0 stepping chips.

 

[Rob Williams]

Well, that's a good point, but can't someone just down clock their RAM if they reach 4GHz? I believe the ratio allows going one step under it, although I'm not sure what that is off-hand. Regardless, it does make sense to take that into consideration. I do believe that very few people have the ambition to go beyond 4GHz though, and even then, it's not going to be stable. So taking things like that into consideration will effect hardcore overclockers only, really.

I'm in such a battle with myself over this ;-)

I'm preparing to do the content, even if I can't make up my mind over things. We haven't had RAM content for a while, and it really shouldn't be ignored, so I feel an obligation to do something. I'd still love to know what tests I could perform that would show the benefits of faster memory though. Adobe Lightroom should be one, but finding others would be sweet also.

 

[Kougar]

Have a look see at the table here: Link

From my understanding (feel free to correct me if I am wrong) there are only three memory ratios to select. 6, 8, and 10. If I buy a Core i7 920, overclock it to 4,000Mhz, I can run my memory at 1200 or 1600... that's going to be a ~5% performance hit right there since both would be CAS 7 timings.

 

[Rob Williams]

By that chart, it looks like 4GHz and a 6x multi would still give a DDR3-1200 clock, so I'm still not sure faster memory is needed. I still believe that DDR3-1600 is a sweet spot though, from both a price perspective and also performance, especially since timings are getting a lot better. I wouldn't mind working on an article specific to this discussion though... perhaps get a DDR3-2000 kit in from someone and use a wide-variety of RAM settings and see the differences in both bandwidth and latency, and then also take a look at application performance.

I'm in talks with a few companies regarding which applications we should be using in our RAM content, so I'm hoping to have a solid plan sooner than later.

 

[Kougar]

I'd be curious to see what programs the memory manufacturers themselves would prefer to see used... I can't imagine them preferring anything non-synthetic? I hope they can offer some real application suggestions to go with those.

Continuing on my line of logic, I'm just stating that it'd be better for someone to buy 1600MHz memory if they can run it at the same timings they would otherwise run a1333Mhz kit when being forced to use 1200Mhz speeds. Especially since 1600MHz (and even 1866MHz) kits are reaching $70 for 6GB now counting rebates.

 

[Psi*]

long winded ?s

The memory testing here proves out what I have experienced with my apps which are all intensive number-crunchers that run from hours to da-a-ays. RAM speed has had little to no impact with any of my systems over the years. It is impressive to see all cores pegged in Task Manager for that time with no disk I/O.

I need lots of RAM and am setting out, again, to build the fastest number cruncher I can w/o going completely nuts. $5K+ would get a pair of X5580s and lots of RAM, but these OC-ed I7s can approach that speed (SPECFP/GEMSFDTD) for under $2K.

My apps do not always use all possible cores as the number of threads are launched depending on problem size. And there is a lot of single threaded processes that have to be completed before that point ... so single core speed is very important & 8 cores might not be fully utilized where all 4 cores would more likely be. So, a I7 975 OCed might finish a project quicker than dual W5580s running at stock simply because all 8 cores of the W5580s would not be maxx-ed because of fewer threads.

Needless to say stability is all important; nothing worse than to have a crash after waiting for 3 or 4 days. Next is processor speed. I have WC and ... ehh ... its ok & I might cannibalize 1 of those machines to place the new guts. My office also gets pretty warm & WCed systems do seem to moderate the CPU temps ... or at least I feel that they do.

I do have a question!!! For max CPU OC of any I7, is the I7 975 *the* chip!?!? For greatest chance at highest CPU clock rate, is 1600 MHz memory all that is needed?

I have followed the threads and almost have all of the timings in my head, but not quite. I will need 12 GB RAM, if that makes any difference & it doesn't seem to.

The 2nd question buried in all of this. The CFP2006 (Floating Point Component of SPEC CPU2006) has some similar number crunchers to what I do. Perhaps something like this could augment your testing? It looks like $800 buys floating point & integer/memory intensive tests ... well, maybe as it takes some study just to figure out what is what on their web site.

So, Rob, I picked you to mention this to as your use of English is ... well, better than any other reviewer I have seen. And, your testing approach is better than most. So, it seems to me that with the growing number of OS's not to mention all of the other little necessities of life such as virus checkers and what not that could be added for other secondary real world tests ... not to mention hardware variations ... that SPEC software could be something you might consider. Nothing better than a good scientifically conducted test to separate the men from the boys, the facts from the baffling & BS. (For instance, just exactly what virus checker has the least impact on my system???)

You also question the need for expensive RAM. I have a dual Opteron 290 system with 16 GB RAM ... Patriot memory. This memory was the cheapest that I could get at the time and it is still good after 3+ years of rarely getting a shut down. THAT I7 system looks to be 6X to 10X faster tho.

 

[Kougar]

No, Core i7 920's (D0 stepping - Box code SLBEJ) are just as good as Core i7 975s for overclocking in most cases.

 

[Psi*]

No, Core i7 920's (D0 stepping - Box code SLBEJ) are just as good as Core i7 975s for overclocking in most cases.

Thanks for replying to my off topic tangent... but just to make sure that we are on the same bit map ... and I have read thru a boat load of reviews, you are saying that the 920 will (or has for some) been over clocked as high as any 975?

I def like the price difference at ~$800 less on Newegg. However, it doesn't take very many hours of run time before that difference could be made up! But, $800 ...

AND, to the topic of this thread memory doesn't make too much difference either. Still looking for maximum smoke ... er, speed.

 

[Kougar]

Specifically D0 stepping Core i7 920's, yes. I can hit 4.3GHz stable with mine... possibly 4.4GHz with voltages I wouldn't care ot use for 24/7 use. 4.2-4.5GHz is about the ballpark for water cooling, 5GHz for subzero cooling.

I've given it plenty of thought, and honestly CAS 6 or CAS 7 1600MHz RAM is all that is needed. If the user plans to overclock to 4.5GHz then 1866MHz CAS 7 would be my suggestion, but not many chips are going to get to a high enough Bclock to justify anything beyond 1600MHz.

 

[Rob Williams]

Hi Psi*, and welcome to the forums!

Needless to say stability is all important; nothing worse than to have a crash after waiting for 3 or 4 days.

This is something to take extremely seriously, because you're right... if the CPU proves unstable after two or three days, you essentially just lost that same amount of time crunching (I assume?). If you are looking for raw speed and a CPU that's going to prove 100% stable, I'd personally spend the money on the fastest CPU Intel has and keep it stock.

Alternatively, if you pick up a Core i7-920 and overclock it, I wouldn't personally go that high... I'd keep it to around 3.5GHz or 3.6GHz max. Even then, you'll want to stress-test any overclock for days using the most robust stress-testers out there in order to be satisfied that your overclock is indeed stable.

I do have a question!!! For max CPU OC of any I7, is the I7 975 *the* chip!?!? For greatest chance at highest CPU clock rate, is 1600 MHz memory all that is needed?

What Kougar said.

The 2nd question buried in all of this. The CFP2006 (Floating Point Component of SPEC CPU2006) has some similar number crunchers to what I do.

We have CPU2006... I just don't know how to use it ;-) If you have any experience, please let me know. We're in the process of revising our CPU testing suite, so this is a good time for me to figure it out. I have a good feeling I'll be doing it in Linux though (which is fine, we desperately need Linux benchmarks).

If worse comes to worse, I'll just e-mail SPEC themselves and ask for advice... because I do agree that we should have it. Both AMD and Intel use CPU2006 to boast about their new processors, so it would make all the sense in the world that we use it as well.

So, Rob, I picked you to mention this to as your use of English is ... well, better than any other reviewer I have seen. And, your testing approach is better than most.

Thanks for the compliments! We put a ton of effort into our methodologies, so it's nice when someone notices ;-)

You also question the need for expensive RAM. I have a dual Opteron 290 system with 16 GB RAM ... Patriot memory. This memory was the cheapest that I could get at the time and it is still good after 3+ years of rarely getting a shut down.

That all depends on the bandwidth, really. Even on non-i7 machines, I couldn't figure out how to tap 5,000MB/s... let alone 20,000 on the i7. I'm REALLY not sure how memory-intensive your particular scenario is, but I assume if you can utilize 16GB of RAM, then the extra bandwidth would prove beneficial.

 

[Psi*]

Trying to stick to "RAM content", there were memory tests that intensely tested the memory system with tests like walk through 1s with all of memory set to 0 and then looks for bit errors ... and many variations of this ad nauseam ... and can take a very long time. The programmers had some knowledge of the hardware and PCB layout as the testing could find crosstalk issues in the physical layout of the memory system. Crosstalk could be the coupling of a signal from 1 trace on a PCB to another causing some kind of memory error. BUT crosstalk could occur because of just a bad board layout near some sockets.

Certainly the tests could find flaky or marginal chips as well. And, as I vaguely remember could identify particular chips so you knew what to replace.

So there are 3 topics relative to memory testing; the memory chips, the sticks or modules, and the mother board.

Now that we have gotten past my own self interests about my build, I remember that this kind of memory testing is what 1st came to mind. Clearly an exhaustive RAM test of this nature would take a long time given GBs of RAM, but very fast processors ... so no telling. It looks like there might be some programs available for today's systems.

Testing might even need to be focused on the mother boards and keep the same set of chips ... the "gold" set(s) if you will. Or the other way, have "gold" performance mother board(s), and change the chips. Either way, probably some obvious economy in there some where driven by $$ & time. And, I would be happy with some kind of gray scale ... good luck figuring that out.

And, who knows, you might get half way into this & truly prove that memory bandwidth makes for nice testimony and that is about all. Or, maybe there are some mobos that might have ample OC features in BIOS, but really do suck (not to get too scientific or anything) at actual capability because of hardware issues.

Last, about *my* stuff, all of my programs have scaled directly with CPU speed ... so my assumption is that CPU/ALU is the limiting "channel". Although not all of the programs peg the CPUs. But, not all number crunchers are created equal or are based on the same solving methods. Some programs might use 80% of the CPUs (or cores) which does cause one to wonder what is the neck of that funnel? Memory channel, inefficient programming, or .... ... So for me, I guess fast memory may be sort of like religion for some people; they "go" just in case.

Until I find out otherwise (from you???) I am going to buy the cheapest CAS 6 or CAS 7 1600MHz RAM, thanks Kougar, mediocre memory that I can find. I will keep 1866 CAS7 for the next build hopefully later in the year if the economy allows.

 

[Rob Williams]

Whoops. I did kind of get off the memory topic there.

I have to be honest here... the tests you describe are severe overkill for our goals. While the results could prove interesting, testing for things like crosstalk sounds more like something that should be tested in a proper lab, with equipment we certainly don't have. I really don't think crosstalk is a huge issue, and if it is, I'd like to be corrected. Memory is more stable than ever, and even when overclocking I'm not sure issues like this would be introduced.

I've tested many stock-speed kits before and have never received errors even after long tests (8 hours). The tests you describe would no doubt take days, and in all seriousness, it would make no sense for us to dedicate so much time to something very, very people would be interested in (we do have limited PCs we can work with).

The most important factor we'd want to take a look at is simply where the bandwidth can be put to good use, and what real benefit can be caused from tighter latencies. Some companies sell DDR3-1333 7-7-7, while others sell DDR3-2000 7-7-7... it'd be our goal to find out exactly why that faster memory is worth looking into.

Regarding your personal scenario, are you using common applications, and is there any way to possibly "benchmark" it with the same workloads over and over?

Last, about *my* stuff, all of my programs have scaled directly with CPU speed ... so my assumption is that CPU/ALU is the limiting "channel".

Sadly, I think that goes for most any application. That's the problem :-/

 

[Psi*]

I wasn't going to add anymore to this thread, bu-u-ut I cannot resist.

I have finally just run Everest on my Supermico H8DC8 e/w dual Opteron 290s. It is fascinating that memory latency is between 59 ns & 60ns. This is with the cheapest possible RAM I could get at the time .... Patriot DDR, ECC, registered, unbuffered, 166 MHz, 2.5, 6, 6, 7.

Needless to point out that with a server board there are very limited OC capabilities. I did just try Nvidia's ntune program ... what can I say? I had time on my hands & my mind was wandering plus I just waited a few days (128 hours) for a model to run.

To answer your question about common applications ... "well, not really". They are commercial apps. but quite expensive @ > $60K. And, the DVDs don't even play music video.

I am looking into some freeware number crunchers that may be similar. Oft times there is a university behind such software & they could be multi-threaded. So a benchmark could be setup running standard models with these programs to compare various h/w setups.

Good benchmarks with these programs would be contingent on problem size & type. It would be great if I could devise just 2 test models; one memory pig but fast simple solution (meaning more scalable with CPU speed) and, the other might have a lot of memory access overhead (meaning a more complex model). More about that on another time.

Back to everest ... FPU Julia is 4760 on all 4 cores on H8DC8. So you can see how I am very interested in more CPU speed.

And, back to Nvidia's ntune ... I blindly jumped in fully committed & set the program up for auto-tweaking, or what ever it is called. Came back a while later & the machine was BSOD-ed. I wasn't really surprised. In attempting to make slight adjustments, manually, I became quite adept in shorting the BIOS reset pads. I could muster an ~9% over clock so my 128 hour run became 123 hour. ehhhh! Not worth the gamble for stability. There is no option of voltage adjust and any tweaking of the RAM was immediate BSOD.

BTW memory bandwidth is 8.40 GB/s per Sandra.

What does this all mean? I don't really know maybe you guys can tell me. I think that it means I need faster CPU like a i7 clocked as high as possible.

 

[Rob Williams]

To answer your question about common applications ... "well, not really". They are commercial apps. but quite expensive @ > $60K.

How on earth does a piece of software cost $60,000? I could see a site license being up there, but a single application?! And here I thought 3ds Max was expensive! Given it's a commercial application, can you share what it is? I'd love to better-understand what could require so much memory and benefit from both huge bandwidth and low latencies, because I've yet to see a great example.

Not worth the gamble for stability.

No sir... not at all. That's what I was hinting at earlier. Imagine running a render for four days and then it BSOD hours before the render was to finish. I can't even picture how aggravating that'd be.

Obviously a faster CPU is going to benefit you, especially if the scenario is heavily multi-threaded. In our Core i7 tests, the HyperThreading feature benefited a LOT of our workstation apps, especially 3D Rendering and to a degree, also our video rendering. I just posted a thread the other day where I'm looking for suggestions and the like for upgrading our CPU testing machine... maybe you could have a look? It's not memory-related per se, but I'd love to hear some recommendations from someone who actually works with such hardcore scenarios day in and day out.

Not to go too far off the intended path here, I wish it were a simple task for me to include both consumer and true workstation benchmarks, but it's a matter of acquiring said applications and also scenarios, because what we have now is a little bit lacking. We're fortunate to have a copy of 3ds Max, but even the model we uses there is lackluster... it's included on the samples disc! I wish there were a resource of free but fully robust projects out there for us to benchmark with... I think our results would greater help the professional if we were able to get some.

</tangent>

Back to your problem though, like I said, it's REALLY hard to say whether faster memory is going to improve much, but at this point it seems like the CPU is the biggest roadblock. If you have the money, though, I'd recommend just buying what you can... a dual-socket Xeon (Nehalem) machine with 12GB of DDR3-1600 RAM would simply be stellar for what your doing (again, as long as your scenarios are truly multi-threaded, and for $60K, they freaking better be).

One a dual-socket motherboard, you're likely to be able to run even more than 12GB, but I'd start off with 12GB (6GB to each CPU essentially) and see if more is needed later on.

 

[Psi*]

haha ... well there are a few packages that I know the price with certainty.
www.ansoft.com www.cst.com www.remcom.com to name 3 3d electromagnetic field solvers. Their primary programs can predict with good accuracy everything from how antennas for communication with satellites & antennas embedded in notebooks & antennas in cell phones next to the human will work to creating a video of an electric charge moving through a PCB & connector.

And, this is just the electrical side of things. www.ansys.com www.flomerics.com have software for calculating mechanical dynamics to "computational fluid dynamics" ... this is how Apple, Dell, HP & the rest figure out how many vents & what kind of fans go into their enclosures. They also predict how molten plastic flows in a mold for computer housing to connector housings.

Soooo, all of that geeky stuff requires a lot of geeky people around the planet writing multi-threaded software (a $10K option from some & almost old hat now a days), distributed network processing (another $10K option from some), to ... make sure you read about MPI clusters that most of these companies have implemented ... no idea how much that costs!

When business was really good, I had 3 of the field solver with several options. "Just tools in the tool box" as they all have their better features. One has a 3D repositionable human body ... necessary for figuring out how/where to put antennas with best range amongst other things.

The SPEC software touches on one of these solver techniques, but those 1st 3 companies actually represent 4 or 5 solution methods AND each use the CPU & memory quite differently. I am happy with just the multi-threaded stuff.

Yes, I have been drooling over the a dual w5580 system that I might build for ~$5K. But not all and even not many projects require that intensity ... I have learned. I believe that the I7 975 is equivalent to "a" w5580. The I7 975 is about $700 less (@ Newegg) and 4 cores are a great thing and not all problems will use all 4 cores. Some, because of smaller size, only setup maybe 2 threads & half the I7 would be used. That would be a big waste with dual w5580s.

Yes, I build my machines rather than pay the extra to Dell. One must be diligent and quick about getting it all together and built ASAP in case there are issues. In that case ... gotta love newegg! The $$ that doens't go into this build (not building dual w5580 system) will go into the next system or software.

I don't know what I could do to help, but I would at least offer my $.02 worth. I get into all of the current PC technologies when I need to upgrade. I am there now, unfortunately the economy has slowed that down but it does give me more time to study it all for the best bang. I will try to keep on the freeware number crunchers to come up with suitable problems.

I think since I am a system builder, it is why I OC. I have an old dual Athlon water cooled system that has been OC-ed for .... well for ever. For the past 3 years, since the dual Opteron box, it has been the best darn email & report writing machine ever! Still running Win2K, I'm too cheap to upgrade it and I am afraid. But, I not afraid of nor unfamiliar with OC-ing.

That's why I have been polluting this thread with questions & comments quite off the original topic about the I7 975 mix with what kind of memory that might have the most likely combination of maximum CPU clock speed. Well, I thought it to be a little related & if the forum were more heavily traveled like some of the engineering or auto forums I would have been more careful, because I sure would have been flamed long before now.