DDR-3000 C12: you have to bin a lot of ICs to get ones with the right voltage/performance characteristics for that kit. Same reason why the more expensive CPUs are also the faster (in MHz numbers or cores) than the cheaper ones.
True. But you can get DDR 2666 with CL10 for about 100€, so a set with an 7% shorter access time (higher "PI" as Ian insists on calling it), and only a 11% lower transfer rate for about a fifth of the price. The 500$ kit seems to be exclusively for those who don't have to work for their money, or maybe those who are hunting records as a hobby.
The very top of the line always is extremely expensive, and - when it's the result of extreme binning - has to be in order to limit demand to the miniscule supply available.
Exactly, they have to test the ICs individually with those tester kits and bin them for speed. I just find it amazing that a chip that is designed for say, 1600 C11 at 1.5v has the potential to run 3100 C12 with 1.65v, that's nearly double its rated clock speed with a mere 0.15v bump in voltage.
When it comes to memory, over the years I have tried to read up on different reviews and look at benchmarks in an attempt to understand when it is better to run tighter timings/lower MHz as opposed to looser timings/higher bandwidth. I'm sure it is a case by case basis, but was wondering if the always knowledgeable and helpful Anandtech commenters could give me a quick, dummy terms, explanation of when tight timings or clockspeed is better? Looking at your graph, it shows the C7 1866 through C10 2666 all having the same performance index score, but what situations do those different timings/MHz become better/worse? I hope this isn't too in depth of a question.
I don't know if this analogy is correct, but I'm seeing it as if RAM was a race car on a track, high bandwidth/loose timings would mean your car travels faster, but has to do more laps around the track to complete. Tight timings/lower bandwidth means the car travels slower but doesn't have to do as many laps to complete. If I am correct on this, at what point does less laps trump traveling faster?
As a side note, I am looking to build a Haswell desktop in Jan/Feb. It will have one GPU (probably one of the R9's) and more than likely a 2x8gb RAM kit. My usage would very roughly be 70% gaming, 25% rendering in 3DS Max and using some Adobe programs, 5% or less video encoding. I'm looking for help in deciding what to look for in this scenario, but also to finally have a better understanding of how these settings affect different workloads.
Ian has a pretty in depth article on this subject, look back thru the archives.
Memory prices haven't increased as much as I thought, if at all... I bought a 1600 4x4GB Patriot kit last year, precisely around this time, it was like $55 or something with a discount (I remember not seeing any better deals around Black Friday). My first exposure to Patriot too but it's worked out well.
I'm not so sure about that, I bought a 32gb 1866 G.Skill Ares kit in December of last year for $109.99 (I'm looking at the invoice) and now the same kit is $283.49 (newegg). That's a huge price increase in my book.
You want my memory scaling article on Haswell: http://www.anandtech.com/show/7364/memory-scaling-... There are 26 different combinations of MHz / CAS there, from 1333 C9 to 3000 C12, representative of many 2x4 GB Hynix MFR kits available to purchase today.
Did you used to have Hyper memory on Nehalem by any chance? That combination had issues above 1.75 volts. I have not encountered any issues running memory up to 1.75v on Sandy Bridge, Ivy Bridge or Haswell. All my chips just work. Otherwise 1.65 volt kits wouldn't be selling as well as they do, and overclockers wouldn't be pushing 1.8v on air / 2.0+ volts on liquid nitrogen.
Yes, because extreme overclockers care so much about longevity. Ian, you should make an effort to get out of the extreme OC and corporate PR echo chambers once and a while. It is really having an effect on the quality of your articles.
Expanding upon this point: the data in Anandtech articles is always top notch, but it is becoming more and more obvious that there are two tiers of reviewers when it comes to delivering insight. Anand, Brian, Ryan, and Jarred write good conclusions based on their data and don't care about any blowback from the manufacturers. Dustin and Ian seem beholden to the manufacturer's PR departments and just parrot whatever talking points they're given. It's really disappointing.
And who exactly do you think is going to be interested in a kit like this, other than overclockers? The review fit the product and the target audience. It's not for general users, never will be, and doesn't need to be reviewed as if it were.
Patriot has changed the ICs on this kit without changing the SKU. I have two of the 2x4GB kits that only have 8 chips on a single side of the PCB and none on the other, and use Hynix H5TQ4G83MFR 4Gbit ICs (the same ones that are on those DDR3-3000+ kits) and clock accordingly. One kit I bought but took back was like these in the review, double-sided sticks with 16 chips per stick (8 per side) and using a relatively new IC, Hynix H5TQ2G83DFR, which can't clock as high as H5TQ2G83CFR unfortunately.
With the demand for MFR seemingly strong, and other companies other than Patriot going after 2400MHz and up, I guess going to CFR was more a financial choice.
Companies seem rather reluctant to tell me which ICs they use, and popping a heatspreader off is no mean feat nowadays, with accidents happening regularly: http://forum.hwbot.org/showpost.php?p=207472&p...
That's compounded by the fact that sometimes the IC # is removed and replaced with the company name over and over. Any suggestions?
Ian, whatever happened to low latencies? I'm currently running a 1600MHz CL7 kit which I bought 2+ years ago, but I only ever see CL9 (and up) kits in reviews or for sale these days.
The reason is that defining latency as a multiple of clocks is rather silly with a large range of clock speeds available concurrently. What your CL7 means is that you have a latency of 4.38 ns (7/1600MHz). The fastest latencies in other clockings available are:
You missed the part where they asked for low latency 1600 and you quoted a 1600 at CL6 without saying where it's from. Like they said, most 1600 kits come at around CL9 which is around 5.63ns. This matters somewhat when Intel CPUs such as the i7 4770K are rated at 1600, any higher and you're running out of spec.
Not sure how I "missed" that, it doesn't say anything about a 1600 kit at CL9 in the question : " I only ever see CL9 (and up) kits in reviews " Well, most kits in reviews and announced sales are probably not 1600 at this point in time. In the review above you see 5 kits at 2400+, with only a single 1600 kit thrown in for completeness. So I assumed that the original poster was expecting DDR3 2400 to also come with CL7. Sorry if that assumption was incorrect.
The quoted CL6 kits are "OCZ Reaper HPC Edition" (OCZ3RPR1600C6LV4GK) and "Super Talent Chrome Series" (WB160UX6G6). I think both are actually discontinued, because you can buy a 2400 CL9 set and just run it at 1600 CL6. As shown above, you could even buy a 2400 CL10 set and get a little lucky and still run it at 1600 CL6 (same latency as the tested 1866 CL7)
So sure, DDR3 1600 kits are rarely sold with very low latencies today, that's because low-latency kits are validated and sold at higher frequencies. This does not matter very much, since all kits come with a JEDEC setting to run 1600 initially, and everybody who knows he needs better latencies can lower them by hand to the actual achievable latency. Kits sold as 1600 are really mainly for people looking for cheap memory. Which is fine, as most reviews show little to no relevant gain from faster memory for most tasks anyways.
No, I wasn't expecting 2400 to come in CL7, but thanks for the assistance. My point is really that kits which qualify as low latency like many of those in your list, seem to be in very short supply and/or very expensive today. I don't think this was the case a couple years ago when there was higher availability.
The ICs that could run at those latencies comfortably have long since been discontinued. They were mostly 1Gbit chips (128MB) made by Elpida and Powerchips, so the biggest sticks you could get out of them were 2GB. Most memory these days are 4GB and 8GB sticks and the ICs used can't run low latencies at speeds of 1600.
Honestly I thought that for a long time, too. But Ian's work is extremely technical and thorough, and once you get into the intricacies of it, there's a lot to learn. I've been spending a lot of time delving into the effects of high speed memory and while 90% of the time it's not important, those weird 10% corner cases can be very compelling.
Playing around with memory can be very interesting in general, and a lot of users swear by high speed memory because it just *seems* smoother. Every build and every system is essentially a game of moving bottlenecks around, and there's some value in being able to take memory speed out of that equation.
And yet the charts show that there is no difference for most use cases. Good data is a great thing to have. However, you shouldn't throw that data out the window and make conclusions based on emotions like you've done in your second paragraph. (Ian is just as guilty of this, so you're in good company.)
I don't know why people just don't accept that there is a market for high speed RAM kits that often has very little to do with 24/7 "real world" usage and move on if it doesn't interest them.
This seems like a pretty good option. I've been a custom builder for the last years and I have always thought that it's really painful to keep track of all the different components. I've tried a couple of different solutions like Evernote, but started using a new startup called Unioncy the other day. Found it quite helpful since they track warranties for you. Might be worth a try if you are building a lot like me.
While this review is technically interesting up to a point, it reinforces a view I've had of the tech press for a while now: there is far too much attention on high end halo products which have norelevance to most buyers. In this case, it appears we're embarking on a review of massively expensive options for ram when the existing tests show absolutely no performance difference whatsoever between these high priced kits and ddr3 1333 ones. Triple the price for an extra 1fps? Recommended!
Any idiot spending hundreds of dollars on ram can afford to replace what he doesn't like. Where are the reviews of budget motherboards, ram kits, processors? The only people who really need to know what the latest $700 processor running on a $400 motherboard with $200 of ram and a $600 graphics card can do can afford to replace what they think isn't good enough and are probably upgrading every 6 months anyway.
You know who *needs* reviews? The guy replacing a $5-600 build from 3 or 4 years ago with the same price, and is faced with hundreds of choices. He can't afford to buy the wrong part and the parts he's choosing from are far more superficially varied and difficult to evaluate.
Needless to say, if you want to find out if $500 card x is significantly faster than $500 card y from 6 months ago, a hundred sites will tell you. Comparing a $100 processor? nope.
The graphs in this article and the last one tell us all we need to know about high end memory: it makes no difference. Please move on!
Indeed, or for those of us that found girls, moved out, got older, changed hobbies, just realised that running benchmarks all day is a waste of life or found that actually the world doesn't end if you don't upgrade your PC every 6-12 months.
There is a need for some sites that analyse how the current $60-$200 GPUs compare with those of 5 years ago, same for CPUs etc. Big market for that kind of info but unfortunately all we get is "this sites for enthusiasts noob!" well thanks but I'm still an enthusiast but now I have a mortgage or I'm only earning half what I was 5 years ago.
The info I get from Anandtech I can get anywhere......
In the conclusion you should add one bar to one of your charts... a bar where the RAM is at 1600 but the cpu is clocked just 100MHz higher, to really highlight how little impact memory speeds have on performance compared to even a tiny cpu clock speed boost.
Ian can you give a little more information about the size of your CPU computer benchmarks specifically the grid size on the finite difference problems. In my experience memory bandwidth plays a large role in the speed of the computation. There are many HPC applications that have memory as the bottleneck and I'm wondering if your problem size is small enough that it is being efficiently handled by the cache and the ram speed isn't making much of an influence. In know in my own CFD code going from 1600 to 1866 showed an almost linear speed up.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
48 Comments
Back to Article
julandorid - Monday, November 18, 2013 - link
Thanks for the review, but what exactly "featured review" means?IanCutress - Monday, November 18, 2013 - link
That's a little tagline we can attach to the front page articles when they're on the top.Wall Street - Monday, November 18, 2013 - link
I think that it is the opposite of capsule review. A featured review is a.k.a. a full review.TemjinGold - Monday, November 18, 2013 - link
Whoa... why is the 2X4 by GSkill $520?IanCutress - Monday, November 18, 2013 - link
DDR-3000 C12: you have to bin a lot of ICs to get ones with the right voltage/performance characteristics for that kit. Same reason why the more expensive CPUs are also the faster (in MHz numbers or cores) than the cheaper ones.ShieTar - Tuesday, November 19, 2013 - link
True. But you can get DDR 2666 with CL10 for about 100€, so a set with an 7% shorter access time (higher "PI" as Ian insists on calling it), and only a 11% lower transfer rate for about a fifth of the price.The 500$ kit seems to be exclusively for those who don't have to work for their money, or maybe those who are hunting records as a hobby.
DanNeely - Tuesday, November 19, 2013 - link
The very top of the line always is extremely expensive, and - when it's the result of extreme binning - has to be in order to limit demand to the miniscule supply available.Gen-An - Tuesday, November 19, 2013 - link
Exactly, they have to test the ICs individually with those tester kits and bin them for speed. I just find it amazing that a chip that is designed for say, 1600 C11 at 1.5v has the potential to run 3100 C12 with 1.65v, that's nearly double its rated clock speed with a mere 0.15v bump in voltage.sf101 - Monday, December 9, 2013 - link
If you want 2400 guaranteed out of the box you pay the premiums.most of the 2133 mhz black momba sticks could also do 2666mhz @ 10-13-10-30-2t but your voltages may vary.
And more than likely some of that is because of individual IMC tolerances per cpu.
Franzen4Real - Monday, November 18, 2013 - link
When it comes to memory, over the years I have tried to read up on different reviews and look at benchmarks in an attempt to understand when it is better to run tighter timings/lower MHz as opposed to looser timings/higher bandwidth. I'm sure it is a case by case basis, but was wondering if the always knowledgeable and helpful Anandtech commenters could give me a quick, dummy terms, explanation of when tight timings or clockspeed is better? Looking at your graph, it shows the C7 1866 through C10 2666 all having the same performance index score, but what situations do those different timings/MHz become better/worse? I hope this isn't too in depth of a question.I don't know if this analogy is correct, but I'm seeing it as if RAM was a race car on a track, high bandwidth/loose timings would mean your car travels faster, but has to do more laps around the track to complete. Tight timings/lower bandwidth means the car travels slower but doesn't have to do as many laps to complete. If I am correct on this, at what point does less laps trump traveling faster?
As a side note, I am looking to build a Haswell desktop in Jan/Feb. It will have one GPU (probably one of the R9's) and more than likely a 2x8gb RAM kit. My usage would very roughly be 70% gaming, 25% rendering in 3DS Max and using some Adobe programs, 5% or less video encoding. I'm looking for help in deciding what to look for in this scenario, but also to finally have a better understanding of how these settings affect different workloads.
Sorry for the wall of text!!
Impulses - Monday, November 18, 2013 - link
Ian has a pretty in depth article on this subject, look back thru the archives.Memory prices haven't increased as much as I thought, if at all... I bought a 1600 4x4GB Patriot kit last year, precisely around this time, it was like $55 or something with a discount (I remember not seeing any better deals around Black Friday). My first exposure to Patriot too but it's worked out well.
hp9000 - Friday, November 22, 2013 - link
I'm not so sure about that, I bought a 32gb 1866 G.Skill Ares kit in December of last year for $109.99 (I'm looking at the invoice) and now the same kit is $283.49 (newegg). That's a huge price increase in my book.IanCutress - Monday, November 18, 2013 - link
You want my memory scaling article on Haswell:http://www.anandtech.com/show/7364/memory-scaling-...
There are 26 different combinations of MHz / CAS there, from 1333 C9 to 3000 C12, representative of many 2x4 GB Hynix MFR kits available to purchase today.
Franzen4Real - Monday, November 18, 2013 - link
Oh great!! Thank You!! Not sure how I missed that in the first place.... Bigtime thanks for all of your time and effort on these tests!!djscrew - Tuesday, November 19, 2013 - link
A (very) basic rule of thumb: 1 Cas Latency = 1 Command Rate = 1 Bump in MHz (1333/1600/1866). With the bump in MHz considered more valuable.dingetje - Monday, November 18, 2013 - link
1.65 volt and higher ?so the haswell memory controller doesnt have a problem with voltage higher than 1.5 or will it get fried eventually ?
IanCutress - Monday, November 18, 2013 - link
Did you used to have Hyper memory on Nehalem by any chance? That combination had issues above 1.75 volts. I have not encountered any issues running memory up to 1.75v on Sandy Bridge, Ivy Bridge or Haswell. All my chips just work. Otherwise 1.65 volt kits wouldn't be selling as well as they do, and overclockers wouldn't be pushing 1.8v on air / 2.0+ volts on liquid nitrogen.dingetje - Tuesday, November 19, 2013 - link
that it runs great i dont doubt....was just wondering wether on the long term the hawell memory controller will have issues (seeing the spec is 1.5)dingetje - Tuesday, November 19, 2013 - link
^haswell (typo)im running a 1.35v (@1.5v) crucial kit myself on haswell by the way
mfenn - Tuesday, November 19, 2013 - link
Yes, because extreme overclockers care so much about longevity. Ian, you should make an effort to get out of the extreme OC and corporate PR echo chambers once and a while. It is really having an effect on the quality of your articles.mfenn - Tuesday, November 19, 2013 - link
Expanding upon this point: the data in Anandtech articles is always top notch, but it is becoming more and more obvious that there are two tiers of reviewers when it comes to delivering insight. Anand, Brian, Ryan, and Jarred write good conclusions based on their data and don't care about any blowback from the manufacturers. Dustin and Ian seem beholden to the manufacturer's PR departments and just parrot whatever talking points they're given. It's really disappointing.Gen-An - Tuesday, November 19, 2013 - link
And who exactly do you think is going to be interested in a kit like this, other than overclockers? The review fit the product and the target audience. It's not for general users, never will be, and doesn't need to be reviewed as if it were.Gen-An - Monday, November 18, 2013 - link
Patriot has changed the ICs on this kit without changing the SKU. I have two of the 2x4GB kits that only have 8 chips on a single side of the PCB and none on the other, and use Hynix H5TQ4G83MFR 4Gbit ICs (the same ones that are on those DDR3-3000+ kits) and clock accordingly. One kit I bought but took back was like these in the review, double-sided sticks with 16 chips per stick (8 per side) and using a relatively new IC, Hynix H5TQ2G83DFR, which can't clock as high as H5TQ2G83CFR unfortunately.IanCutress - Monday, November 18, 2013 - link
With the demand for MFR seemingly strong, and other companies other than Patriot going after 2400MHz and up, I guess going to CFR was more a financial choice.Companies seem rather reluctant to tell me which ICs they use, and popping a heatspreader off is no mean feat nowadays, with accidents happening regularly: http://forum.hwbot.org/showpost.php?p=207472&p...
That's compounded by the fact that sometimes the IC # is removed and replaced with the company name over and over. Any suggestions?
Gen-An - Tuesday, November 19, 2013 - link
CFR would have been preferable to DFR, which as you review and this one show, doesn't like going higher than 2600: http://www.techstation.it/articoli/patriot-viper-i...chekk42 - Monday, November 18, 2013 - link
Ian, whatever happened to low latencies? I'm currently running a 1600MHz CL7 kit which I bought 2+ years ago, but I only ever see CL9 (and up) kits in reviews or for sale these days.joos2000 - Monday, November 18, 2013 - link
Lower latencies doesn't yield the same performance returns as upping the clock frequencies, that's why.http://www.anandtech.com/show/7364/memory-scaling-...
ShieTar - Tuesday, November 19, 2013 - link
The reason is that defining latency as a multiple of clocks is rather silly with a large range of clock speeds available concurrently. What your CL7 means is that you have a latency of 4.38 ns (7/1600MHz). The fastest latencies in other clockings available are:1066 CL7 => 6.56 ns
1333 CL7 => 5.25 ns
1600 CL6 => 3.75 ns (But only on 2GB kits)
1866 CL8 => 4.29 ns
2133 CL9 => 4.22 ns
2400 CL9 => 3.75 ns
2666 CL10 => 3.75 ns
2800 CL11 => 3.93 ns
3000 CL11 => 3.67 ns
So as a matter of fact, all kits tested in this review, except for the ADATA ones, have shorter latencies than your own set.
Gigaplex - Tuesday, November 19, 2013 - link
You missed the part where they asked for low latency 1600 and you quoted a 1600 at CL6 without saying where it's from. Like they said, most 1600 kits come at around CL9 which is around 5.63ns. This matters somewhat when Intel CPUs such as the i7 4770K are rated at 1600, any higher and you're running out of spec.ShieTar - Tuesday, November 19, 2013 - link
Not sure how I "missed" that, it doesn't say anything about a 1600 kit at CL9 in the question :" I only ever see CL9 (and up) kits in reviews "
Well, most kits in reviews and announced sales are probably not 1600 at this point in time. In the review above you see 5 kits at 2400+, with only a single 1600 kit thrown in for completeness. So I assumed that the original poster was expecting DDR3 2400 to also come with CL7. Sorry if that assumption was incorrect.
The quoted CL6 kits are "OCZ Reaper HPC Edition" (OCZ3RPR1600C6LV4GK) and "Super Talent Chrome Series" (WB160UX6G6). I think both are actually discontinued, because you can buy a 2400 CL9 set and just run it at 1600 CL6. As shown above, you could even buy a 2400 CL10 set and get a little lucky and still run it at 1600 CL6 (same latency as the tested 1866 CL7)
So sure, DDR3 1600 kits are rarely sold with very low latencies today, that's because low-latency kits are validated and sold at higher frequencies. This does not matter very much, since all kits come with a JEDEC setting to run 1600 initially, and everybody who knows he needs better latencies can lower them by hand to the actual achievable latency. Kits sold as 1600 are really mainly for people looking for cheap memory. Which is fine, as most reviews show little to no relevant gain from faster memory for most tasks anyways.
chekk42 - Tuesday, November 19, 2013 - link
No, I wasn't expecting 2400 to come in CL7, but thanks for the assistance. My point is really that kits which qualify as low latency like many of those in your list, seem to be in very short supply and/or very expensive today. I don't think this was the case a couple years ago when there was higher availability.Gen-An - Tuesday, November 19, 2013 - link
The ICs that could run at those latencies comfortably have long since been discontinued. They were mostly 1Gbit chips (128MB) made by Elpida and Powerchips, so the biggest sticks you could get out of them were 2GB. Most memory these days are 4GB and 8GB sticks and the ICs used can't run low latencies at speeds of 1600.djscrew - Tuesday, November 19, 2013 - link
RAM reviews seem more and more pointless by the day.Dustin Sklavos - Tuesday, November 19, 2013 - link
Honestly I thought that for a long time, too. But Ian's work is extremely technical and thorough, and once you get into the intricacies of it, there's a lot to learn. I've been spending a lot of time delving into the effects of high speed memory and while 90% of the time it's not important, those weird 10% corner cases can be very compelling.Playing around with memory can be very interesting in general, and a lot of users swear by high speed memory because it just *seems* smoother. Every build and every system is essentially a game of moving bottlenecks around, and there's some value in being able to take memory speed out of that equation.
mfenn - Tuesday, November 19, 2013 - link
And yet the charts show that there is no difference for most use cases. Good data is a great thing to have. However, you shouldn't throw that data out the window and make conclusions based on emotions like you've done in your second paragraph. (Ian is just as guilty of this, so you're in good company.)Gen-An - Tuesday, November 19, 2013 - link
I don't know why people just don't accept that there is a market for high speed RAM kits that often has very little to do with 24/7 "real world" usage and move on if it doesn't interest them.Ewram - Tuesday, November 19, 2013 - link
I would have liked an AMD example on IGP performance as well, but oh well...Ewram - Tuesday, November 19, 2013 - link
Forgot to finish my post with: Great review anyway!johnnyfoxes - Tuesday, November 19, 2013 - link
This seems like a pretty good option. I've been a custom builder for the last years and I have always thought that it's really painful to keep track of all the different components. I've tried a couple of different solutions like Evernote, but started using a new startup called Unioncy the other day. Found it quite helpful since they track warranties for you. Might be worth a try if you are building a lot like me.Hairs_ - Tuesday, November 19, 2013 - link
While this review is technically interesting up to a point, it reinforces a view I've had of the tech press for a while now: there is far too much attention on high end halo products which have norelevance to most buyers. In this case, it appears we're embarking on a review of massively expensive options for ram when the existing tests show absolutely no performance difference whatsoever between these high priced kits and ddr3 1333 ones. Triple the price for an extra 1fps? Recommended!Any idiot spending hundreds of dollars on ram can afford to replace what he doesn't like. Where are the reviews of budget motherboards, ram kits, processors? The only people who really need to know what the latest $700 processor running on a $400 motherboard with $200 of ram and a $600 graphics card can do can afford to replace what they think isn't good enough and are probably upgrading every 6 months anyway.
You know who *needs* reviews? The guy replacing a $5-600 build from 3 or 4 years ago with the same price, and is faced with hundreds of choices. He can't afford to buy the wrong part and the parts he's choosing from are far more superficially varied and difficult to evaluate.
Needless to say, if you want to find out if $500 card x is significantly faster than $500 card y from 6 months ago, a hundred sites will tell you. Comparing a $100 processor? nope.
The graphs in this article and the last one tell us all we need to know about high end memory: it makes no difference. Please move on!
jabber - Tuesday, November 19, 2013 - link
Indeed, or for those of us that found girls, moved out, got older, changed hobbies, just realised that running benchmarks all day is a waste of life or found that actually the world doesn't end if you don't upgrade your PC every 6-12 months.There is a need for some sites that analyse how the current $60-$200 GPUs compare with those of 5 years ago, same for CPUs etc. Big market for that kind of info but unfortunately all we get is "this sites for enthusiasts noob!" well thanks but I'm still an enthusiast but now I have a mortgage or I'm only earning half what I was 5 years ago.
The info I get from Anandtech I can get anywhere......
Shadowmaster625 - Tuesday, November 19, 2013 - link
In the conclusion you should add one bar to one of your charts... a bar where the RAM is at 1600 but the cpu is clocked just 100MHz higher, to really highlight how little impact memory speeds have on performance compared to even a tiny cpu clock speed boost.jabber - Tuesday, November 19, 2013 - link
Buy whatever matches best with your motherboard and GPU colour scheme I say.ShieTar - Wednesday, November 20, 2013 - link
Of course. And never overclock your memory when there is a full moon.D1RTYD1Z619 - Tuesday, November 26, 2013 - link
or if you have pickles in your fridge.rmh26 - Tuesday, November 19, 2013 - link
Ian can you give a little more information about the size of your CPU computer benchmarks specifically the grid size on the finite difference problems. In my experience memory bandwidth plays a large role in the speed of the computation. There are many HPC applications that have memory as the bottleneck and I'm wondering if your problem size is small enough that it is being efficiently handled by the cache and the ram speed isn't making much of an influence. In know in my own CFD code going from 1600 to 1866 showed an almost linear speed up.UltraWide - Tuesday, November 19, 2013 - link
Did you have a chance to remove the heat-spreaders to see which ICs are in these? I am assuming Hynix MFR?Gen-An - Tuesday, November 19, 2013 - link
Hynix DFR actually. 2Gbit ICs (256MB each) so the same size as CFR but with nowhere near the overclockability.