Page 3 of 5 FirstFirst 12345 LastLast
Results 33 to 48 of 72

Thread: Best RAM make for NF7-S?

  1. #33
    HEXUS.social member Agent's Avatar
    Join Date
    Jul 2003
    Location
    Internet
    Posts
    19,185
    Thanks
    738
    Thanked
    1,609 times in 1,048 posts
    Originally posted by Austin
    Believe it or not I'm one of those boring types that like to keep things in stock (as much as possible) ... esp voltages.
    Wheres the fun in that ?

    The vDD mod is easy, the vDimm is the one im really not looking forward to.
    Debating on if i should lap the NB while im at it (i suppose its somthing to do this weekend)
    Quote Originally Posted by Saracen View Post
    And by trying to force me to like small pants, they've alienated me.

  2. #34
    Banned
    Join Date
    Jul 2003
    Location
    House without a red door in Birmingham
    Posts
    1,595
    Thanks
    0
    Thanked
    0 times in 0 posts
    Originally posted by KAIN
    Oh dear

    Keeping stuff at stock :'(

    That's not good
    More of in spec than stock. Running a TbredB XP1700+ @ XP2800+ speed is not necessarily running it out of spec IF you can keep to a normal voltage like 1.65v. Even though that CPU usually uses only 1.50v to run at XP1700+ the core is capable of much higher speeds and designed for higher voltages. Another example is a Rad9500PRO, Rad9600PRO, GF4TI4200 or GF3TI200 ... each of those uses technology that is far more capable than they are set to at stock ... so o/c'ing them to a higher speed (minus adding voltage)and in some cases worrying the top of their range is not 'abusing' the hw as such, they are still running in a way they were designed to and the technology is capable of in a usual way (or diff card / model). I'm not trying to say I don't o/c nor that o/c'ing is a bad thing ... just that I'm very careful when it comes to voltages as they throw off a LOT of heat and seriously stress any given part in a way it wasn't designed for. Plus I really hate any source of noise, added voltage = heat so not good for quiet(er) rigs.

    It's adding voltage to RAM and mobos' chipsets which worry me most ... as I said I am pretty boring when it comes to experimenting and pushing the boundaries.
    Last edited by Austin; 17-10-2003 at 02:10 AM.

  3. #35
    Senior Member Russ's Avatar
    Join Date
    Jul 2003
    Posts
    5,201
    Thanks
    11
    Thanked
    69 times in 44 posts
    Originally posted by Austin
    Yeah I bought most of my RAM from Crucial in the past but recently the seem a little behind in latency, o/c'ability...

    Believe it or not I'm one of those boring types that like to keep things in stock (as much as possible)..
    hypo_ _ _ _ _ fill in the blanks

  4. #36
    Official Member luke313's Avatar
    Join Date
    Jul 2003
    Location
    West Mids
    Posts
    921
    Thanks
    0
    Thanked
    0 times in 0 posts
    hypopottamus??

  5. #37
    Banned
    Join Date
    Jul 2003
    Location
    House without a red door in Birmingham
    Posts
    1,595
    Thanks
    0
    Thanked
    0 times in 0 posts
    Originally posted by Russ
    hypo_ _ _ _ _ fill in the blanks
    So I can only rec to others what I do myself? That's dictatorship. I give people the info and they can make up their own mind if they want to push things. As for (your beloved) Crucial they use CL3.0 while practically every other manu's PC3200 uses CL2.5. If you quote my full post(s) you'll also find I say Crucial overcharge (not like Corsair do) so the likes of Adata and TwinMOS are much better options, whether o/c'ing or not. It also makes sense to have RAM that is likely to be able to handle the faster speeds should you wish to o/c in the future ... like somebody did with their TwinMOS PC2700 when 400mhz(& FSB) ended up being the norm.

  6. #38
    ERU
    ERU is offline
    Who in Arda is Ilúvatar ERU's Avatar
    Join Date
    Aug 2003
    Location
    Caerdydd
    Posts
    1,878
    Thanks
    104
    Thanked
    9 times in 8 posts
    I totaly agree with the Dual Channel certified and low latencycomment - it not worth the extra. If you got loadsa ££'s go for it - if like mosy of us you havent then dont bother.

  7. #39
    Member
    Join Date
    Aug 2003
    Location
    Dooawn Souf
    Posts
    60
    Thanks
    0
    Thanked
    0 times in 0 posts
    Personally my feeling is overclocking full stop is not good for hardware. I have had graphics card I kept at stock voltage but overclocked die far sooner than cards overclocked but with increased voltage.

    Heat is as much of a problem as voltage, same as stressing components overclocking is a problem as well. The components are made to run at a set speed and they will die sooner whether you add extra voltage when overclocking or not.

    All my personal opinion, as long as I keep it at a reasonable temp I'm happy to add the voltage that's my test

  8. #40
    Senior Member Russ's Avatar
    Join Date
    Jul 2003
    Posts
    5,201
    Thanks
    11
    Thanked
    69 times in 44 posts
    does twinmos come with life time garantue?

  9. #41
    Banned
    Join Date
    Jul 2003
    Location
    House without a red door in Birmingham
    Posts
    1,595
    Thanks
    0
    Thanked
    0 times in 0 posts
    Originally posted by KAIN
    Personally my feeling is overclocking full stop is not good for hardware. I have had graphics card I kept at stock voltage but overclocked die far sooner than cards overclocked but with increased voltage.

    Heat is as much of a problem as voltage, same as stressing components overclocking is a problem as well. The components are made to run at a set speed and they will die sooner whether you add extra voltage when overclocking or not.
    Running something physically faster will add more heat but no where near as much as adding voltage to it will, esp as most o/c'ers do both. I forget the techie term but using more voltage than a part is technically designed for will almost certainly cause it long term damage, you effectively burn out the data pathways. All parts must have a degree of headroom in them, added to which many of the low to mid-range parts use the same technology as the high end ones (mostly just underclocked) and are often clocked far lower than they (and their technology) are capable of simply to make the high end offerings shine or to fill demand in a cheap and easy way. So using voltages above the design of the part is a gamble while running it at a faster speed is ... well ... a much smaller gamble so long as you're careful.

    It does depend what you wish to o/c but there are plenty of parts which are hugely more capable than they are set to by default, primary examples being P4C 2.4ghz, XP1700+ to XP2000+ TbredB, XP2500+ Barton, GF3TI200, GF4TI4200, Rad9500PRO, Rad9700 and Rad9800. O/c'ing per-se does not damage nor age your component, so long as you follow simple rules and keep away from the absolute maximum stable o/c. Voltage is certainly the most damaging thing (not that I'm saying run a Barton @ 3ghz as that would be like pushing 9v thorugh it).

    I can't find any specific info on TwinMOS' warranty, some sources state it's lifetime while others state 'unknown' ... the minimum it could have is 12 months AFAIK as that's std practice on the part of any retailer (and presumably supplier). Of course if you o/c (even tweak the timings) you invalidate your warranty, lifetime or not. Crucial is top pick for server and mission critical operations because of its well known quality, reliability, warranty and customer service. I wonder what current pricing is like ...

    http://www.jjas-it.com/acatalog/Computer_Memory.html

    Pricing:

    CPUcity TwinMOS PC3200 CL2.5: 256MB £34, 512MB £68.
    CPUcity TwinMOS PC3700 CL2.5: 256MB £53, 512MB £105.
    (Komplett about the same pricing)

    Crucial PC3200 CL3.0 £35, 512MB £63.

    Dost my eyes deceive me? Seems Crucial have adjusted their prices, they were way more expensive when last I looked. Okay so it is CL3.0 (we haven't seen such poor latency since PC133 days) but latency is way over-rated esp for Dual Channel ... not sure what diff CL3.0 vs CL2.5 would make though? Unless this is just down to the normal downward slide of RAM pricing and retailers selling TwinMOS haven't got the cheaper batches in yet it would seem Crucial want to compete again.

  10. #42
    Member
    Join Date
    Jul 2003
    Location
    52:35:52N - 2:08:57W
    Posts
    107
    Thanks
    0
    Thanked
    0 times in 0 posts
    Ok, the deed is done. 1Gb TwinMos + 2500 Barton is on its way

  11. #43
    HEXUS.social member Agent's Avatar
    Join Date
    Jul 2003
    Location
    Internet
    Posts
    19,185
    Thanks
    738
    Thanked
    1,609 times in 1,048 posts
    Originally posted by Austin
    I forget the techie term but using more voltage than a part is technically designed for will almost certainly cause it long term damage, you effectively burn out the data pathways.
    Electromigration
    Quote Originally Posted by Saracen View Post
    And by trying to force me to like small pants, they've alienated me.

  12. #44
    Member
    Join Date
    Aug 2003
    Location
    Dooawn Souf
    Posts
    60
    Thanks
    0
    Thanked
    0 times in 0 posts
    Just wondering Austin, have you done electronics or ever done a course on processor design?

    I did both during my degree, an unfortunatley any electronics run outside of guaranteed spec even if they are capable will decrease the life expectancy. You damage the pathways anyhow...

  13. #45
    Banned
    Join Date
    Jul 2003
    Location
    House without a red door in Birmingham
    Posts
    1,595
    Thanks
    0
    Thanked
    0 times in 0 posts
    It's not what you know it's who you know (or is that meant to be the other way around?). I understand what you're saying and I understand you're coming from a more generic or perhaps paper-orientated position of knowledge. It's just not as simple as you seem to think (not that I claim to know it all to any degree LOL).

    Real world example the Rad9500, 9500PRO, 9700 and 9700PRO. ATI design their DX9 part, get all the technology and archy sorted and produce their top card the 9700PRO. However they need to fill the more lucrative lower sectors. They decide to adapt their 9700PRO design to fill these lower sectors rather than go through loads of R&D specifically tailoring 'new' cards to these sectors (esp with the unknown factor of nVidia's 'new' DX9 part waiting in the wings). So they set clocks lower which is also useful to minimise wastage (binning), this is the 9700 (nonPRO) which sits nicely behind the 9700PRO. Next they cut the (expensive) memory bandwidth of the 9700 (256bitDDR) by moving to the commonly used 128bitDDR, that's the 9500PRO and again fills the gap nicely. Finally they want (roughly) an entry level DX9 part for which they cut half the pipes from (again minimises wastage from 8 piped 9500PRO-9700PRO).

    There you have it, a full lineup with offerings from (near) entry level right up to the high end. All use the same 0.15mu design with BGA RAM but basically all the cores are identical. They covered themselves if they got cores unable to run at 9700PRO speed as they could be used in any of the other cards. However as it turned out even early cores seem very happy at 9700PRO speed regardless of where they ended up, in fact many are capable of much higher speeds than even the 9700PRO is set to. Of course there's always a possibility of having a poor part that just runs at its limit even in 9500-9700 cards ... but talk to anyone who tried o/c'ing and you'll find it's very rare not to exceed the 9700PRO core clock of 325mhz.

    The same applies to the GF4TI although the 4200 used TSOP and 6 layers rather than the BGA and 8 layers of the 4400/4600 ... still it didn't hold it back much as it still almost always exceeded 4400 speed. The same is seen in CPUs, it's not like AMD R&D intending to produce XP1700+ right up to XP2800+, they just set the CPUs where they need to depending on demand. You research and develop and then underclock (and undervolt) the part in order to give you the full range from entry level to high end. Of course there's parts in the range that are not going to achieve the high-end part's speed, some of it is down to preventing wastage.

    So it is (IMHO) clear that if you pick your part you can o/c without any negative effects on the part in question, esp if you drop back a little from the maximum stable speed and of course avoid things like unusual voltage. Another bonus is since we're dealing with low to mid end products they are cheap to replace should the worst happen.

    Running things out of spec does not necessarily reduce the life expectency of a part, it's not as simple as getting 80% of its lifespan because you o/c and as covered above you can o/c many parts and not reduce the life at all (as the technology is designed for a faster speed than the part is set to) ... anyway I'm sure you know all that.

  14. #46
    Panzer Division Marduk PanzerKnight's Avatar
    Join Date
    Jul 2003
    Location
    Surrey
    Posts
    555
    Thanks
    6
    Thanked
    0 times in 0 posts
    my mushkin lv 2 pc 3200 runs fine on this mobo and with the barton 2500+

  15. #47
    HEXUS.Metal Knoxville's Avatar
    Join Date
    Jul 2003
    Location
    Down In A Hole
    Posts
    9,388
    Thanks
    484
    Thanked
    442 times in 255 posts
    • Knoxville's system
      • Motherboard:
      • Intel X58
      • CPU:
      • Intel i7 920
      • Memory:
      • 2GB DDR3
      • Storage:
      • 1TB
      • Graphics card(s):
      • ATi HD3450
      • PSU:
      • Generic
      • Case:
      • Cheap and nasty
      • Operating System:
      • Vista 64
      • Monitor(s):
      • 24" LG LCD
      • Internet:
      • Virgin Media 20mb
    Originally posted by Austin
    latency is way over-rated esp for Dual Channel ... not sure what diff CL3.0 vs CL2.5 would make though?
    Latency is not over-rated, A jump from 3 to 2.5 is noticable as is a jump from 2.5 to 2. I know i remember noticing it, even in dual channel

    Fair enough its not the most important aspect of a system, but it does make a difference during real usage not just in benchmarks.

  16. #48
    HEXUS.social member Agent's Avatar
    Join Date
    Jul 2003
    Location
    Internet
    Posts
    19,185
    Thanks
    738
    Thanked
    1,609 times in 1,048 posts
    In most cases on the NF2 , its better to have have normal timings and go for the higher FSB speed.
    Quote Originally Posted by Saracen View Post
    And by trying to force me to like small pants, they've alienated me.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •