OCZ 4200EL: Tops in Memory Performance

by Wesley Fink on 10/13/2003 9:33 PM EST
POST A COMMENT

45 Comments

Back to Article

  • kgs - Friday, January 02, 2004 - link

    ya, I just have to read everything 4 times.
    Now I just really need to know:

    I have read all the reviews, and I'm gonna copy-cat as much as possible, because there is a known means to an end. I will be preparing to tackle any cooling issues if req'd at any time. I have 2 questions: (1)How much do I need for my specific purpose, and if you can (2), should I get the specific processor.

    I am upgrading because here in Toronto, using Rogers hi-speed cable internet, I have recently been scouring the newsgroups. These contain upwards of 10, 000, 30,000, or 100,000 messages in each group. I get all the headers in each of some groups each day. I mark about half a dozen total for download, and then finally combine and decode them, currently using OE. My old 192 MB, cusl2-c w/ celeron600(I know ) just crashes, hangs, waits, etc.. to excess.

    Q1
    How much EL4200 does a serious homegrown OE nntp userneed: (2x256), or (2x512)? I may not have the chance to play around. I can only do things once at a time.

    I am upgrading to an ASUS P4C800E-deluxe. I plan on also getting the 2.6C box cpu, because it seems like a sweet spot, and is not the baby in the C series. I assume the 2.4, 2.6, 2.8, and 3.0 behave the same. From reviews I assume the 3.2 behaves somewhat differently, maybe more ocable. It is also a lot more money. If recommended I could also get the 2.4C(~$???), 2.8C(add~$50), or 3.0C(unlikely)(add ~$150). I just tried to fill in this 2.4C(~$???) value above from a www site to get my facts all in a row, and found the 2.4 and 2.6 similarly priced. I have read a couple times in reviews that the 2.4C has been used and they said it was used because it was very overclockable. I assumed this would be true similarly about any of the C b/t 2.4C-3.0C. I also assumed the articles I was reading may be older, and that the 2.4C was probably the sweet spot when the article was written, and they would have achieved similar(and therefore better) with a 2.6C.

    Q2
    Any comments on the 2.4C vs 2.6C question.
    Reply
  • hepp - Friday, January 02, 2004 - link

    kgs,

    No, they are single sided and single banked.
    It is still dual channel though, since you get 2 sticks.
    A 512 kit is two individual modules that have been tested together.
    Two individually purchased sticks should perform much the same but have not been tested together.

    512MB PN-OCZ533512ELDC-K
    is a kit consisting of 2 Single bank 256MB modules.

    1024MB PN-OCZ5331024ELDC-K
    is a kit consisting of 2 double bank 512MB modules.

    Hope this helps
    hepp
    Reply
  • kgs - Wednesday, December 31, 2003 - link

    I am easily confused.
    I am refering to the "Dual Channel Kit" w/ copper heat spreader. Not 2 seperately purchased units.
    Are ther differnt Dual channel kit EL4200 512 total (I have written 2x256). I assume this kit is double sided ram.

    And 2x256 will give slightly lower benchmark scores since they are single bank.

    4) There are single bank (2x256) kits as well.
    There is a review of the 2x256 kit at legitreviews.

    Reply
  • hepp - Monday, December 29, 2003 - link

    kgs,

    1) Do you mean 2x256 as opposed to 2x512?
    It is likely that 2x256 would allow slightly higher FSB all else equal. The reason would be that 2x512 requires more voltage.
    But since individual modules vary 280 is not guranteed with 2x256 or 2x512.
    And 2x256 will give slightly lower benchmark scores since they are single bank.
    1x512 would allow higher FSB but would give worse performance.

    2) 2.6 or 2.8 would need less FSB for same CPU speed, so you probably would not need 4200 modules with those unless you have great cooling.
    2.4 @ 280 = 3360Mhz
    2.6 @ 259 = 3360Mhz
    2.8 @ 240 = 3360Mhz

    If you have a great CPU and cooling you could turn that around and say that you can get better overall performance with those at 280.
    2.6 @280 = 3640Mhz
    2.8 @280 = 3920Mhz

    3) Don't think that either of those will affect your o/c potential since you can lock the PCI/AGP bus speed.
    Your mobo,PSU and cooling will...

    4) There are single bank (2x256) kits as well.
    There is a review of the 2x256 kit at legitreviews.

    5) Bonus answer to the question you did not ask.
    OCZ are now speed binning their best Hynix chips with a view to released PC4500 modules. This means that the best chips will not be used for 4200EL so it is likely that recent modules will not perform as well as the ones reviewed here.

    Good luck
    hepp
    Reply
  • kgs - Sunday, December 28, 2003 - link

    I am only slightly aware of the inching problems inherent in achieving the best in trio performance. That is why I am so thrilled about this article I'll never forget it.

    In reading the last chart it seems no generalizations can be made, so not that I am finicky, but I have some older components that give me concern about seeking for these numbers blindly.


    Please, now that I have found this comments section after searching around like some chicken, I have 4 questions for anyone regarding the 3-3-4-7 @2.85V @280/560/1120 TIMING, V , FSB VARIATION I can expect:

    1)
    USING 512MB, not 1024MB, but using the same slots.

    2)
    when I buy the 2.6C, or 2.8C, rather than the 2.4C

    3)
    Will these AGP and PCI components I want to keep using have any effect on this pursuit that has been demonstarted is possible:
    Radeon 64MB DDR VIVO AGP4x
    SBLive 5.1 pci
    v.90 USR/3comm voice 2976 pci
    Pioneer 16x DVD 116 IDE cable player
    IBM Deskstar DTLA-307030 30 GB 7200 ATA-100 (data storage only)

    N.B. A WD 36 GB Raptor SATA will soon be o/s drive , and I may get a +/- DVD/CD r&rw

    4) is there only one EL4200. e.g double sided, ...
    I am here in Toronto, ON, if that makes any diff
    Reply
  • hepp - Monday, November 10, 2003 - link

    Could you please, pretty please, do a comparision between betwen OCZ 3500 LE at 5:4 and OZ 4200 EL at 1:1.

    It would be really interesting to see how they compare at 250, 260,270 and 280.

    Br
    hepp

    Reply
  • Anonymous User - Tuesday, October 21, 2003 - link

    I read your latest review of OCZ's 533 memory.
    You said it was the fastest ram you have tested to date.
    How does it compare with RAMBUS PC1066 or one of there other very fast modules.

    Greg Kelly
    Reply
  • Anonymous User - Monday, October 20, 2003 - link

    do you plan on testing this memory on an AMD platform? Reply
  • Anonymous User - Thursday, October 16, 2003 - link

    I find it hilarious that so many guys just wait for OCZ to come out with anything just so they can bash it. I mean, come on, for the last year OCZ has dramatically improved their service and has become a leader in ram technology. Are there other good companies? Yes. Do they have ram that is 500+ ddr... some. Are they just as expensive..yes! Do they have the same customer service? Not on your life.
    Give me a nice p4 chip that will do 280, and some OCZ4000 or 4200 so I can run 280 fsb and 1:1 and I would be a happy man... Oh wait, I do have that!

    PJ
    Reply
  • billyzbear - Thursday, October 16, 2003 - link

    Is that a "hand picked" 2.4c? That is a real nice oc for that chip. You may have hit the max for that chip. I thought at one time you guys had a unlocked Intel chip is the only reason I asked, thanks for the reply.
    Reply
  • Wesley Fink - Wednesday, October 15, 2003 - link

    As detailed in the review, the CPU is our standard for high-FSB memory testing, a 2.4C (800FSB). This CPU runs at default CPU voltage to about 290 on the P4C800-E, and can go to a bit over 300 (1200FSB) at 1.6V. At 280, the CPU was running near 3.4GHz at DDR560 memory at default CPU voltage. Reply
  • billyzbear - Wednesday, October 15, 2003 - link

    Sorry if I missed it but what type cpu was used, one that has unlocked multi? I don't have my pen and paper but at 1-1 that seems like a crazy oc. What was the final cpu speed? Reply
  • Anonymous User - Wednesday, October 15, 2003 - link

    I don't care about timings. People who obsess over latency are, again, uninformed.

    And I don't hold anything against OCZ or anyone else for marketing this stuff. They are businesses, and businesses exist for the sole purpose of making money. Making money at any point in the DRAM chain is very hard right now, so if they can entice suckers to pay outrageous sums for the products then that is what they will do. But supposedly unbiased reviewers are supposed to point these things out. Some sites did. AT did not.
    Reply
  • Anonymous User - Wednesday, October 15, 2003 - link

    #25
    why don't you try to overclock the best ddr400 to ddr560. you can loosen the timing. you can give the memory 3.3+ voltage. and you know what i doubt if u can get any close to ddr500. and btw i believe 3.5ns and 5ns makes a hell lot of difference. ohh, i forgot thej are remarked 5ns chips. and one more thing basic micro economics evri company tries to deferenciate their products so it can sell them on a premium price. just like mercides - u can find better quality cars much cheper - and there is still people paying a hell lot of money to drive a mercides. product defferenciation!
    Reply
  • Anonymous User - Wednesday, October 15, 2003 - link

    #25
    why don't you try to overclock the best ddr400 to ddr560. you can loosen the timing. you can give the memory 3.3+ voltage. and you know what i doubt if u can get any close to ddr500. and btw i believe 3.5ns and 5ns makes a hell lot of difference. ohh, i forgot thej are remarked 5ns chips. and one more thing basic micro economics evri company tries to deferenciate their products so it can sell them on a premium price. just like mercides - u can find better quality cars much cheper - and there is still people paying a hell lot of money to drive a mercides. product defferenciation!
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Personally im pretty damned impressed with it. OCZ are fast approaching DDR600 - something Id never have thought would happen by the end of this year. Only 12 months ago everyone was limited to around DDR470 - thats great work in my view. Its showing progression. Early DDR400 had crappy timings too - given time thsi stuff will mature and we will see the latencies start to fall.

    PCs live in a world of gradual improvement. You have to look at the 5% improvement in perspective. 5% of a lot is still a lot afterall. You all seem to be complaining that youre not seeing sudden huge leaps in performance. Well thats fine, youre all welcome to sit there with your pcs, not change a thing for 2 years then buy a whole new rig and get your 50% speed improvement. Thats not what memory and other components like this are made for. Theyre made for people who want the best and are willing to pay for the best. Afterall you only get a 5% improvement going from a radeon 9700 pro to a 9800XT on most high end systems - you dont see people bitching about that. All those little 5%'s add up. 5% faster cpu, 5% faster gfx, 5% faster memory gives you a 15% speed boost. Sure its not cheap, but the best never is.

    Open your eyes and stop getting stuck in yesterday. The days of low latency memory are gone and wont be coming back any time soon. For any sort of progression to be made people need to stop crying about timings and start looking at performance figures. In my view a 5% increase just from swapping a couple of sticks of ram is good.

    This pointless whining about manufacturers is starting to grate a bit too. Look what OCZ have done in the past year and tell me if theyre producing rubbish good? On the weight of the 3700 Gold alone they couldve ended the year happy, but they seem to be going hell for leather to break DDR600 too, so I say good luck to them.

    Nice review Wes, and well done OCZ.
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    #27 if you have issues email me oczguy2@ocztechnology.com im here to help.

    #26 I have 3500Gold running 2-2-3-5 street racer, enhanced and CPC enabled at 2.65V on the IC7 and max3...will that do ;-) It also ran turbo and CPC at 233fsb 2.5-3-3-7 2.65V.

    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    forgot to mention, with upped voltages (+.02) running CAS 2.5-3-3-7 Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    My OCZ 3700 Gold wont even do 233mhz, with a I865PE motherboard Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Could you tell me for this memory (and for future "universal" memory) the maximum speed at CAS 2? I ask this because the agressive PAT settings (street racer and F1 on my ABIT IC7-G mobo) will only run at CAS 2. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    #24

    What do you expect? Companies push overclocked memory as something "new"? When in fact each module gets pushed up with voltage to get the specs they are saying. It's a joke to sell the same speced DDR400 modules as something else. I'm just suprised Samsung allows remarking of falsely advertised products. To say you are getting true DDR533 is the biggest joke. Considering they are OVERCLOCKING a product and selling it as something else.

    Couldn't you say that they are remarking chips and selling them at a faster mark? I wonder if you could get away with sueing a company based on false advertising a product
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    21 is right. These RAM reviews are presented simply with the pointless synth benches and there is no analysis or interpretation. Since the vast majority of people simply don't know any better, they see "fastest" and assume their system will be boosted into the stratosphere. As anyone who actually bothers to test these modules properly knows, that is false. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    http://www.ocztechnology.com/products/memory/OCZ+E...

    Voltage is listed at 2.8V, this is board specific, some boards will run at DDR533 at 2.7V or less.2.8v is stated for compatiblity.

    The ram will be available from today from http://www.atacom.com they will show stock as soon as it arrives.

    oczguy2
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    #21, your comments are foolish at best when it's a fact that there are PLENTY of people that buy 875P boards. Christ, get a clue. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    "For the most part the 865PE comes within 2 - 5% of the performance of the 875P, which makes the 875P a tough sell. For the majority of users, we would strongly recommend the 865PE because of the significant cost savings. If you're the type of user that must have the fastest thing on the block and will not rest without the knowledge that you do in fact have the fastest thing on the block, then the 875P is for you, but for everyone else, the 865PE is more than sufficient."

    http://www.anandtech.com/chipsets/showdoc.html?i=1...

    Why doesn't Anand just do all the reviews? :-( The above clip is exactly what these memory articles need-sanity and perspective.
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    "You are correct that it makes no difference at all if you only run Microsoft Word and surf the web with dial-up."

    Ironic statement, since Quake 3 is less system intensive these days than Word.
    Reply
  • Wesley Fink - Tuesday, October 14, 2003 - link

    OCZ will have the full info up in about an hour according to the email I just got. OCZ will post a link in these comments as soon as the page is up. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    #16 just because your broke doesn't mean all of us are....for me it will be awesome I run an a64 on a prommie and run at high 1:1 speeds this stuff will allow me to get to 560mhz fsb easy with my cooling which to me and many other overclockers is absolutely awesome....just because you can't afford good components doesn't mean we're all broke. And why does it matter what wes said he showed the performance increase in the benchmarks if you don't think it's worth the money then don't buy it.... Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    wes is this the memory your reviewing....
    http://www.ocztechnology.com/products/memory/OCZ+D... really appreciate an answer because I'm looking for the stuff you reviewed with the 2.5-4-4-7 timings and can't find it....
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Come on man, you're smarter than that. People buying this stuff aren't programmers or content creators, they are gamers. So if you can prove that this RAM offers more that a few % advantage, then it would be nice if you did it. Simply saying that it does carries little weight. So it's the highest 1:1. So what? Does turning a dirt country lane into a superhighway let more cars through if there wasn't any traffic to begin with? This stuff will likely cost over $400 for 1GB. So I'd like to think it offers more than a dent in the wallet. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    awesome review wes...thanks for the good work...i was less then an hour away from ordering the corsair xms4000pro and you changed my mind to ocz (i have ocz el 3200 now)..... Reply
  • Wesley Fink - Tuesday, October 14, 2003 - link

    #13 -
    OCZ says it will be available beginning 10/14, which is today's date.
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    I might have missed something, but when will it be available? Screw all the crap said above. I want to try some out, thats what it is all about. Reply
  • Wesley Fink - Tuesday, October 14, 2003 - link

    #11 - This memory performed at 2-3-3-5 at DDR400 as shown in the review. If you choose not to buy anything but official Jedec DDR400 memory then that is your business, but I would appreciate it if you read enough of the article to get your facts correct.

    #4 and #7 - We can also show you that in REAL APPLICATIONS in Content Creation that the Intel integrated Graphics are just as fast as a 9800 PRO XT. To state from those REAL APPLICATION tests that the 9800 PRO XT is just as fast as Intel on-chip graphics would be a lie - just as your supposition is only a portion of the true picture. Memory is only one part of the performance puzzle, but it does make a difference. It is not serving our readers to overhype it's importance, but it is also not serving readers to pretend it makes no difference at all. You are correct that it makes no difference at all if you only run Microsoft Word and surf the web with dial-up.

    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    I agree above. How can this be better then using DDR 400 with timings a hell of a lot better? Cas 2.5 with a 4-4-7 timing? That's sad. I rather be at 400 with a Cas 2 with 2-2-5 any day.... Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Got to love OCZ ripping off quotes from other companies. "Powered By" ? Oh please................ Once again stealing quotes from other companies seems to be in their deck of cards. What a shame. Plus guys there really isn't such thing as DDR533. It's just overclocked memory! You're a fool if you buy this stuff. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Thanks for the awesome review Wes.

    Tony
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    You crackface, there ARE performance benefits, that's the point. These modules are for extreme users, not some dope on a budget that can't afford an extra $100. No idiot in their right mind should go for DDR533 if they can't spare $100 for a system. Come on now... Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    #5, results? Here's results using ACTUAL APPLICATIONS instead of pointless synthetic benchmarks. If you can give me a reason to pay double the price or more for such minuscule benefits, I'm listening.

    http://www.xbitlabs.com/articles/memory/display/co...
    Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    awesome memory now we can only wait for the "GOLD" version. I bet it will be faster. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Problem is #4, DDR500 isn't useless. You're ignoring results if you can't see that. Reply
  • Anonymous User - Tuesday, October 14, 2003 - link

    Can you guys come up with something a little more useful than Sandra and the 4-year old Q3? Maybe something that actually illustrates how silly and useless this and other DDR500+ really is? Reply
  • Zuni - Tuesday, October 14, 2003 - link

    The dupe entries were addressed.

    L8r
    Reply
  • Anonymous User - Monday, October 13, 2003 - link

    Pulldown menu has DDR400 test results page linked 4 times so everything after that doesn't work properly Reply
  • wicktron - Monday, October 13, 2003 - link

    great job wes. Reply

Log in

Don't have an account? Sign up now