Dismiss Notice
Attention AVForums app / Tapatalk users
Sadly GDPR means that, from 25th, we can no longer offer access to AVForums via the branded app or Tapatalk.
Click here for more information.

RAM compatibility

Discussion in 'PC Gaming & Rigs' started by rvdbarnes, Oct 20, 2005.

  1. rvdbarnes

    rvdbarnes
    Standard Member

    Joined:
    Mar 7, 2005
    Messages:
    89
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    6
    Location:
    sheffield
    Ratings:
    +0
    morning all

    Im now onto the task of choosing the RAM for my computer im building, but am worried about getting some that is incompatible and wont work, is it easy to get it wrong.
    All i know for sure is that that i want to start off with 2x512mb and that my motherboard accepts DDR400 modules.
    A link for my board is below if that helps.

    http://uk.asus.com/products4.aspx?l1=3&l2=15&l3=148&model=382&modelmenu=1

    Ta

    Rvd
     
  2. Uridium

    Uridium
    Member

    Joined:
    Jan 29, 2005
    Messages:
    14,434
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    North West Leicestershire
    Ratings:
    +1,855
    Just by decent branded memory and you'll be fine.

    Crucial.com's online site has a dropdown menu system that selects the correct memory for your M/Board
     
  3. rooney

    rooney
    Active Member

    Joined:
    Mar 7, 2004
    Messages:
    506
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    31
    Location:
    midlands
    Ratings:
    +33
    i always use corsair value select never been let down yet about £30 per 512 from ebuyer etc
     
  4. Tejstar

    Tejstar
    Well-known Member

    Joined:
    Dec 4, 2002
    Messages:
    22,808
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    London
    Ratings:
    +1,081
    Another vote for Crucial here, they offer guaranteed compatability if you order using their tool.
     
  5. Ethics Gradient

    Ethics Gradient
    Banned

    Joined:
    Jul 29, 2003
    Messages:
    3,681
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    aka Billy Science - Suni ojna Tas
    Ratings:
    +430
    Just go for the fastest cas timings you can get on pc3200 ram ( within a sensible budget ofc ) ..... the amd64's are performance affected by latency. Preference is cas2 or faster --- its down to the way the amd64's work with their onboard memory controller.

    memory timings are more important on amd64's than pentiums - it can make a difference of a 4% performance in gameing for example in farcry.
     
  6. pragmatic

    pragmatic
    Well-known Member

    Joined:
    Nov 9, 2004
    Messages:
    12,150
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,647
    EG can you back up such claims with numbers (without oc'ing just stock)? I'm still not a great beliver in timmings although you can get some coursair mid range stuff that has 2-2-2-3 or something timings from ebuyer, at about £80-90 for a matched pair.

    coursair as mentioned above (i think)
    http://www.ebuyer.com/customer/prod...2hvd19wcm9kdWN0X292ZXJ2aWV3&product_uid=52471

    cheap but branded
    http://www.ebuyer.com/customer/prod...2hvd19wcm9kdWN0X292ZXJ2aWV3&product_uid=97038
     
  7. Ethics Gradient

    Ethics Gradient
    Banned

    Joined:
    Jul 29, 2003
    Messages:
    3,681
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    aka Billy Science - Suni ojna Tas
    Ratings:
    +430

    My preference would be to spend the extra £20 and get the quicker ram.
    I wouldn't be as interested if I was running a different chip - but with the AMD 64's, I think its worth it.

    No overclocking - just stock ram:

    What you have to remember, is that everything works on clock cycles or parts there of - so CL of 2 ( and the other lowered settings ) mean that often data can be transfered across in a single clock cycle - where as slower ram say of CL 3 may require slighly more time ... and therefore skip into the next clock cycle.

    Your system waits for all sort of other things during the process of running applications - so its waiting for memory - access to disk controllers - data to the graphics cards etc etc etc.
    So Latency of memory can be a bottle neck within a system of bottle necks - so intead of giving a 33% or 50% increase in system performance, it gives savings of 1% - 5% in normal application operations ( ie standard daily usage )

    Examples of performance differences

    and if anyone wants a better understanding of what all those 2-2-2-6's etc mean along with how latency and burts rates work and affect the system ...

    Explination of memory timings

    ..... some people may find it rather heavy going
     
  8. pragmatic

    pragmatic
    Well-known Member

    Joined:
    Nov 9, 2004
    Messages:
    12,150
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,647
    Yea, nice graph and does show difference, although 1-3/4% could be attributed to an number of different things happening and/or difference could be attributed to taking the best and worst values from a number of runs(although i see no reason why they would need to do this) or even a single run of each was made and the values were somewhat scewed. To make a point, i ran the fear demo sequence twice and both times there was a difference in frame rate and it was arround this range.

    The theory that it should be faster is sound especialy with the internal memory controler built on the chip (so there is no going via another chip which would pretty much destroy any small advantage given, via either bus screw or other latency destroying process) as is is often with computer a great idea turns out not to work quite as well as it should, usualy down to the multiple layers that are involved.

    When it comes to the man on the street the difference between the cost of cas 3 and 2 memory is pretty much non existant now so even the a small performance improvement could be justified.

    It seems rather strange that everyone and his dog consider the lower timings being better but hardly anyone has done an indepth review of this (especialy when cas 2 cost £200 and cas 3 was £100 for the equivalent size), almost like mass hysteria or old wives tale that everyone takes at face value becuase there told so and it makes scientific sense, even though they haven't seen any number to prove so.
     
  9. Ethics Gradient

    Ethics Gradient
    Banned

    Joined:
    Jul 29, 2003
    Messages:
    3,681
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    aka Billy Science - Suni ojna Tas
    Ratings:
    +430
    I noticed a big difference in performance ( sometimes > 10% ) during synthetic mathematical operations when I used to run in my processors - I no longer oc - but I used to do it with a passion ... hence why theres a fridge compressor sat next to my computer bench at home ;)

    I agree that fluxations in benchmarks occur regularly over the course of operation for an exact same spec.
    I used to do all my benchmarks against a vanilla drive with only the specific drivers needed - and the benchmarks were always run directly after startup before any other applications were on the platform.

    This usually gave a more solid regular benchmark with a close +- differential and an average was taken over n runs. When benchmarks were taken after a machine had be used for a while, those differences were exagerated by other processes, vram / pagefile usage etc

    Whether the benchmarks online I posted were due to fluxations are were an true example or not, they do seem to line up with the maths. On a correctly setup system, it 'should' be faster on paper - simply because it can potentially get data read within a single cycle rather having to wait for a second or third .... and therefore within 1 system clock tick. Over over 2.6Gigacyles per second, that does add up when it shunting to and from memory.

    As I said - this does seem to be more relevant to amd64 chips .... so where in the past, you might have observed little to no difference in real terms on regular pc's, there is more chance that it make a significant enough difference to be noticable over the day to day usage of a machine based on the 64's.
     
  10. pragmatic

    pragmatic
    Well-known Member

    Joined:
    Nov 9, 2004
    Messages:
    12,150
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,647
    I have never been one for benchmarking but when i buy a product i like to research the hell out of the options availible, when it came to memory it always seemed conspicious that there were very few benchmarks and those that were heaverly oc based, and as such they weren't really testing latnecy at all rather just bandwidth of the entire system. Pumping up the system bus gives huge improvements, but this is more due to every component in the system running faster instead of just the memory, and the reviews jumpted to the conculsion that x memory was faster than y becuase of it.

    Obviously back then the price of the memory ment you wanted to get the most out of it and this usualy involved oc'ing which is understandable but they never really answered the question was the 100% increase in price worth the increase given (and also the tests were somewhat pointless as unless you had the same system as them your results would vary widly) and the answer is quite blatently no, your better off buying cheaper ram and motherboard (although still decent branded for stability reasons) and speend as much a you could on the graphics card and cpu. Someone probably made a suitability matrix for which was the best fit, i.e. when the graphics card was more demanding than the cpu.
     
  11. Ethics Gradient

    Ethics Gradient
    Banned

    Joined:
    Jul 29, 2003
    Messages:
    3,681
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    106
    Location:
    aka Billy Science - Suni ojna Tas
    Ratings:
    +430
    When I was overclocking, I was doing specific benchmarks on the ram timings -- then ramping up the frequency and testing to get the lowest latency settings against bus frequency.
    The actual ram timings go up at higher clock rates simply because the ram can't handle the quick timings under such pressure.
    So in effect, you have the same ram timings as cheaper ram - but at a far higher clockrate - should you have tried it with cheaper ram, the latency timings would have to have increased also to be 4,5 or 6 instead of 3 for example.
    So timings are relevant in the equation even with overclocking.

    Your comments on price vs performance increase ( real or imagend ) was relevant a few years ago, when low latency ram was considerably more expensive.
    But due to the uptake of such products, the price gap has considerably shortend in many cases ..... your own examples show this.

    £65.10 for normal 1gig ( 512 x 2 )
    £88.95 for low latency 1gig (512 x 2 )

    Thats quite abit of the 100% mark ;)

    If you are planning on spending £1,000's of pounds on systems and marginal increases in performance are not that much of an issue - fair enough. Its like anything else - if you are on a budget and just want a machine operational, buy the cheapest bits you can to fulfill your need.
    But if you are wanting to squease out extra performance out of a machine to get the most out of it, you have to spend more money on a quality motherboard, ram, etc as well as just looking at the CPU and graphics card.
    The higher quality the rest of your components are, the more noticable the performance of less able parts will be.
    You can put cheap retreads on a car, but don't try going for a high speed burn down the motorway.
     
  12. overkill

    overkill
    Well-known Member

    Joined:
    Nov 6, 2003
    Messages:
    11,842
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Murkeyside
    Ratings:
    +1,205
    Wish you'd said this before! :rolleyes: :D

    I'd have to, on recent experience, agree with EG here. Bought my usual ram, 1 gig of the ever reliable Crucial DDR400, which has a cas latency of 3. After getting a few stutter problems with games (on 'high' settings mind) I saw someone else mention this issue with 64 bit CPU's (and some boards come to think of it!), and who gave some effective benchmarks to back his comments.

    Result, I bought some Corsair XL low latency ram (2-2-2-5). So far, it's made a big difference in performance, both in games, and more surprisingly, using apps as well. I say surprisingly as I thought I wouldn't notice apart from with games! There is quite a big price premium, in that the Corsair is a good £30-35 dearer overall, but it's been worth it so far.

    Only problem now is, that the games think my VGA card can handle as much speed as the CPU and ram! It can't............ :(
     
  13. pragmatic

    pragmatic
    Well-known Member

    Joined:
    Nov 9, 2004
    Messages:
    12,150
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    133
    Ratings:
    +1,647
    At the time faster ram with worse latencies were availible (obviously still are altough 500mhz with cas 2 has appeared recently) and were cheaper than those with the super fast timings, so I always thought it was a false economy, if you were gonna just push the fsb anyway. This is all retrospective as times have changed, oh and the price increase is 35% ish for 1-4% performance increase :rolleyes: but as overkill says hes system is performing much better now, over what was before and a difference in benchmarks does not display this ( i guess actual performance is much more difficult to analayse).

    In truth graphics cards are the only component is selected wisely that have a purley linear price/performance, ofcouse this is depening on application to some extent, but a 7800GTx that costs twice as much as a 6800GT is about twice the performance (and faster than an 6800G SLI config). Cpu on the other hand the performance peaks at a certain point and then stagnates when you compair cost vis performance with the more expensive. For example a £700 cpu is not 7x better than a £100 one (unless you get conned maybe). But there is a median were your paying more for performance while still getting a sizable boost, this is usualy always mid range in pretty much every product. Once you move above the midrange the price hikes quickly and you pay for excustivity, but as this is the AV forum i'm sure everyone on here is pretty aware of that.
     
  14. Joe Pineapples

    Joe Pineapples
    Well-known Member

    Joined:
    Oct 23, 2001
    Messages:
    7,352
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    136
    Location:
    Doncaster
    Ratings:
    +1,134
    Overkill - can i ask what the rest of your system specs are?.
     
  15. overkill

    overkill
    Well-known Member

    Joined:
    Nov 6, 2003
    Messages:
    11,842
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Murkeyside
    Ratings:
    +1,205
    Sure thing Joe:-

    Athlon 64 3400
    Corsair 1 Gig XL ram
    ATI Radeon 9800 pro
    Creative Audigy 2
    ASUS K8V-SE
    Seagate 160gb 8mb cache P-ATA HDD
    Netgear NIC
    Enermax Coolergiant 480 watt 3 fan PSU.
    2 x Sunon 12 volt case fans (front and rear)
    Didn't bother with a 3rd fan as the PSU is a push pull with a blower on the base.
     
  16. Joe Pineapples

    Joe Pineapples
    Well-known Member

    Joined:
    Oct 23, 2001
    Messages:
    7,352
    Products Owned:
    0
    Products Wanted:
    1
    Trophy Points:
    136
    Location:
    Doncaster
    Ratings:
    +1,134
    thanks overkill

    i've been reading this thread with interest, as i'll be making a memory purchase decision myself shortly (as well as 2 other mates). The upgrade will be roughly base around...

    A64 dual core 3800+
    Nforce 4 SLI based board
    7800GT.

    We did come to the conclusion initially that with a budget of around £160 for the memory, we would get more benefit from 2gigs of the cheaper Corsair value select stuff, than 1gig of similar stuff to yours. But after reading your experience, i'm now not sure. Its actually made me look at the memory speed as a possinle bottleneck, rather than the point of view of simply having more ram to smooth things out - though i saw first hand evidence of the latter, on a system very similar to yours, with BF2.

    have you noticed a difference with any other games?.
     
  17. overkill

    overkill
    Well-known Member

    Joined:
    Nov 6, 2003
    Messages:
    11,842
    Products Owned:
    0
    Products Wanted:
    0
    Trophy Points:
    136
    Location:
    Murkeyside
    Ratings:
    +1,205
    I bought the ram for improving HL2 really, but it has also meant a marked improvement in FarCry, and less intensive games such as SP Chaos Theory, and Thief Deadly Shadows which could also run a little rough on top settings. The apps that have speeded up include Paintshop pro (man, that took ages to load/react to changes before!). I will be able to say more once the ram has bedded in longer and more apps/games have been tested.
     

Share This Page

Loading...
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice