OCZ PC3200 Premier

In the past two years, OCZ has moved from relative obscurity to becoming one of the most widely recognized names in enthusiast products. Today, OCZ markets power supplies, heat sinks, memory voltage boosters, and thermal compounds, but OCZ is still best known for their enthusiast memory. Recently, OCZ has also been quite innovative in new product introductions, including products like our current RAM Speed champ, OCZ VX, which requires a very non-standard 3.5 to 3.6V to really shine.

What many forget is that like Corsair, Kingston, Mushkin, and other full-line memory companies, OCZ has a very large line of value products. The Premier Series is a Value line, and the PC3200 Premier is a typical Value product, with a web selling price as low as $121 at mWave.

OCZ uses cheaper packaging for their Premier line than their high-end products, but that fits the target, which is low cost. Like Mushkin, OCZ chose to include heatspreaders on their Value product; in this case, copper-colored heatspreaders. The Premier test DIMMs came as a matched pair of 512MB DIMMs in a 1GB kit.

PC3200 Premier is built with PSC memory chips, a Taiwan company that provides chips for value products from several memory manufacturers.

Specifications

OCZ rates their PC3200 Premier at the more aggressive CAS 2.5 timings at DDR400 and stock voltage.

 OCZ PC3200 Premier (DDR400) Memory Specifications
Number of DIMMs & Banks 2 DS
DIMM Size
Total Memory
512 MB
1GB
Rated Timings 2.5-3-3-7 at DDR400
SPD (Auto) Timings 2.5-3-3-7
Rated Voltage 2.6V

Test Results

The OCZ PC3200 Premier is the first Value RAM we have tested that reaches DDR480, our next test point on the 2.4GHz ratios. Our samples actually booted as high as DDR500, but memtest86, Super Pi and games were not stable beyond DDR480, regardless of voltage.

OCZ PC3200 Premier (DDR400) - 2 x 512Mb Double-Bank
CPU Ratio at 2.4GHz Memory
Speed
Memory Timings
& Voltage
Quake3
fps
Sandra UNBuffered Sandra Standard
Buffered
Super PI 2M places
(time in sec)
Wolfenstein - Radar - Enemy Territory fps
12x200 400 DDR 2.5-3-2-6
2.6V 1T
535.0 INT 2637
FLT 2826
INT 6072
FLT 6025
83 112.4
11x218 436 DDR 2.5-3-3-6
2.7V 1T
538.5 INT 2803
FLT 2969
INT 6481
FLT 6410
82 113.0
10x240 480 DDR 2.5-3-3-6
2.9V 1T
546.6 INT 2901
FLT 3136
INT 6708
FLT 6626
81 114.7
11x240
(2.64GHz)
Highest CPU/Mem Performance 2.5-3-3-6
2.9V 1T
586.0 INT 2976
FLT 3214
INT 7083
FLT 7007
75 124.1
To be considered stable for test purposes, Quake3 benchmark, UT2003 Demo, Super PI, Aquamark 3, and Comanche 4 had to complete without incident. Any of these, and in particular Super PI, will crash a less-than stable memory configuration.

The Premier actually did a bit better than specified at DDR400 with stable 2.5-3-2-6 timings. It is also worth noting that Premier was stable with CAS 2.5 timings all the way to the memory speed limit of DDR480. This is particularly outstanding performance for a memory that actually sells for $121.

Aida 32 is now available as Everest Home Edition, a free download from www.lavalys.com. Everest has been very useful in measuring read/write performance and memory latency.

OCZ PC3200 Premier (DDR400) 2x512Mb Double-Bank
Everest 1.51
CPU Ratio at 2.4GHz Memory Speed Memory Timings
& Voltage
Everest READ
MB/s
Everest WRITE
MB/s
Everest Latency
ns
12x200 400 DDR 2.5-3-2-6
2.6V 1T
5795 1948 47.1
11x218 436 DDR 2.5-3-3-6
2.7V 1T
6122 1998 44.6
10x240 480 DDR 2.5-3-3-6
2.9V 1T
6661 2217 44.3
11x240
(2.64GHz)
Highest CPU/Mem Performance 2.5-3-3-6
2.9V 1T
6668 2710 40.3

With OCZ Premier covering a broader range of memory speeds from 400 to 480, there is more potential to see the true impact of memory speed on performance. With memory timings at a consistent 2.5-3-3 and CPU speed at a constant 2.4GHz, any performance deltas are the result of true performance differences. As we move from 400 to 480 - a 20% speed increase - memory read improves by 15%, write improves by almost 40%, and latency drops by about 17%.

However, these improvements in synthetic memory benchmarks are not reflected by similar improvements in benchmarks with real games. Improving memory speed by 20% - from 400 to 480 - at the same CPU speed improves gaming performance by just 2% to 3%. Clearly, we are getting a better performance increase than this in real-world performance with faster CPU speed. You can see this in the 10% CPU speed increase from 10x240 to 11x240 at the same memory timings. This 10% CPU Speed increase improves gaming FPS by about 8%.

Mushkin EM Series PC3200 OCZ PC3200 Value Series: "Value VX"
Comments Locked

102 Comments

View All Comments

  • Wesley Fink - Tuesday, April 12, 2005 - link

    #52-#55 - There are Part numbers for ALL the memories in the review in the chart on Page 2. OCZ responded in post #44 that the 3-4-4-8 and 2.4-3-3-8 were the same VX memory. The parts are rated at what they can do at standard voltage - not what they can do at 3.0V and above.

    #45 - Continuing your analogy. If a Ferrari (Value VX) were available for the same price as a Chevrolet Cobalt, then I think readers would expect me to include the Ferrari in my review of Value cars - despite the fact the Ferrari might need hard to find racing alcohol (3.0V to 3.5V) to run properly while the Cobalt only needed easy to find unleaded regular gasoline (2.6V).

  • adg1034 - Tuesday, April 12, 2005 - link

    Isn't that what he did with the Value VX? Check out the article...
  • Zebo - Tuesday, April 12, 2005 - link

    How bout the real "value" modules which really are cheap and contain same chips as high end ram? Just not speed binnned but who gives a rats ass for that kind of value.

    Some I can think of right off top off head are:

    Crucial's with micron G's = Ballistix for half price.

    TwinMOS with Winbond VX = OCZ VX for less than half price.

    Kingston VR with Hynix BT = All those high end sparkly packaged 4200 modules for half price.

    And so on.

    Meh not what I was expecting.
  • Hardtarget - Tuesday, April 12, 2005 - link

    good article except for one thing! I wish you had compared generic Samsung OEM ram.

    I don't know about the US but in Canada right now OEM Samsung game is super cheap. you can get 1gig for 100 bucks canadian. It's incredibly popular.

    Would of been a great starting point to the article and good to see how it compares to non oem versions.
  • Ranger123 - Tuesday, April 12, 2005 - link

    I think you need to clear up the issue regarding Corsair not providing memory. Corsair's response indicates why they chose not to supply their Value Select parts, but it doesn't explain why no XMS modules were provided. At least a couple of the XMS kits meet your criteria and I assume that these are the parts they were referring to when they said that OCers should look at their parts that are "geared to performance and overclocking".

    I can see 4 possible reasons for Corsair's refusal to send an XMS kit:
    1. Corsair misread the AT request and thought only the Value Select parts qualified.
    2. Corsair doesn't want anything in their XMS line associated with the phrase "Value RAM".
    3. As others have suggested, Corsair knew their modules wouldn't perform well and decided that no result was better than a bad result.
    4. Corsair is using the same rotating chip trick on these XMS modules that they are on the Value Select, they're just not admitting it.

    Given Corsair's reputation I would assume that it's one of the first two, but if Corsair is resorting to some questionable practices I think the AT community would want to know. Maybe you can contact whoever you deal with at Corsair and get an explanation.
  • Crassus - Tuesday, April 12, 2005 - link

    When I saw the headline of the review I expected the review to be somewhat different, to be honest. The tests in itself are not bad at all, but if you still have the samples I ask you both in my name and the names off all the folks complaning about the voltages to perform at least some tests at Voltages around 2.8 - 2.9 Volts to allow us to see how these RAMs perform at other boards (e.g. nForce3+4) and how far they will go in MHz @1T.

    I, too, have to question the ethics of this review in the light of the recent debate. How do we know that the samples you tested are of the same kind as we can buy, if its up to the manufacturer to send you whatever they see fit? Especially when it comes to !value! parts I doubt it will be too much of a problem to recieve them through other channels who will not be as interested in the outcome of the review as the manufacturer (Retail chains or online shops?). You may even consider buying them yourself anonymously to remove any doubt and sell them off lateron. Shouldn't be too hard with a community like this one or an online auction site.

    Cheers,
    Crassus
  • Backslider - Tuesday, April 12, 2005 - link

    I'd like to comment on my experince with Value and Special Edition type RAM.

    While recenly building a set of PCs, I was reading articles similar to this one. I was convinced that in order to get a good system with future potential of over clocking I would need to buy the more expensive (non-value) RAM.

    I happly paid approx. 50% more for this "Extreme Memory", thinking it would be perfect for a stable system. However, after I built my pair of systems I ran into trouble.

    One of the two systems failed to install Windows. And to make a long story short, I ended up with a bad pair of RAM. I RMA'ed them to and received yet another faulty set of RAM. Finaly, I sent them to the maufacturer for an RMA. This time I received a set that hardly squeek by at stock speed. Overclocking by even 1% means instant BSOD.

    Since this happened, I've been buying Value RAM. (from a different manufacturer) And havent run into any problems yet.

    I also find it interesting that the manufacturer of the faulty RAM that I received, declined to Anandtech's request for test samples. Also, there reasoning was a bit questionable considering that their out of their "XMS" line, I received 4 (pairs) of modules that all performed VERY differently. (1 great, 2 bad, 1 not worth the RMA)

    Thats just my input, based on my experience and nothing more.

    -Backslider
  • Den - Tuesday, April 12, 2005 - link

    #54 (and 52).
    I agree, if you are certain you will never overclock you should just get cheap ram with ok timings at stock speed. Interesting that the timings on the picture that you mention (2.5-3-3-7) match what is in the article here.

    I understand why the articles don't have links to buy it, but it really would help if they would at least include the manufacturer's part/model number, then we could be sure.
  • segagenesis - Tuesday, April 12, 2005 - link

    #53 - Actually clicking on the image you can see the ram is marked differently than the newegg page (wtf!) as 2.5-3-3-7 in the picture but 2.5-4-4-8. Ummm...

    Unfortunately I must be somewhat critical when my eyes spin around an article that gives praise to memory for $115 yet doesnt mention *where*. Let alone just a single link to buy it at any price.

    Now granted the Corsair value runs at 2.5-3-3-7 also and likely has little headroom for people who want more out of less, but I consider it a good deal for those like me who dont overclock memory. *takes a breath* With the OCZ being $148 off the same site vs. $87 Corsair, for stock performance its not exactly "value" to me unless you are an overclocker. More like "value if you take the risk". A damn shame Corsair would not submit samples.
  • Den - Tuesday, April 12, 2005 - link

    re: #52
    http://www.newegg.com/app/ViewProductDesc.asp?desc...
    has timings a little closer but you are right, it sure would be nice if they put actual model numbers in the reviews instead of leaving us to guess.

Log in

Don't have an account? Sign up now