PDA

View Full Version : Double Crucial Ballistix review: PC12800 XMP and PC14400 EPP



K404
08-12-2008, 03:48 AM
We should all know who Crucial are, so i'll skip the really obvious stuff.


Press statement:

Fremont, CA, and Glasgow, UK, July 23, 2008 — Lexar Media, a leading global provider of memory products for digital media, today announced immediate availability of Crucial Ballistix® PC3-10600 (DDR3-1333MHz) and PC3-14400 (DDR3-1800MHz) high-performance memory modules. These new modules expand on the award-winning Crucial Ballistix DDR3-1600 and DDR3-2000MHz product offering. As part of its commitment to the PC enthusiast and gaming community, Lexar Media continually seeks to bring new memory offerings to market and enhance its product lines. For complete product details on the Crucial Ballistix line-up, visit: www.crucial.com/ballistix, www.crucial.com/uk/ballistix or www.crucial.com/eu/ballistix.

Specific Crucial Ballistix modules, such as the DDR3-1600MHz, -1800MHz, and -2000MHz modules, have enhanced performance profiles within the module programming that when paired with a supported platform, provide users with an option for easy setup and configuration instead of a manual process that may require trial and error or advanced system knowledge. The modules also support manual configuration in the event the user wants to maintain full control.

____________________________

Box and RAM:


http://img360.imageshack.us/img360/9797/box6zj3.th.jpghttp://img183.imageshack.us/img183/605/dscf0040yg8.th.jpg (http://img360.imageshack.us/my.php?image=box6zj3.jpg)

The packaging is common for both sets- quietly understated and minimalist.

The RAM sticks themselves are a bit more flashy- bright yellow heatspreaders with a chromed aluminium trim.


http://img503.imageshack.us/img503/5900/dscf0014pn9.th.jpg (http://img503.imageshack.us/my.php?image=dscf0014pn9.jpg)http://img230.imageshack.us/img230/5791/dscf0069fn5.th.jpg (http://img230.imageshack.us/my.php?image=dscf0069fn5.jpg)

It would be bad manners not to show all four sticks together....


http://img503.imageshack.us/img503/7639/dscf0051cu3.th.jpg (http://img503.imageshack.us/my.php?image=dscf0051cu3.jpg)


This review will show how the DDR3-1600 and DDR3-1800 parts square off against each other and hopefully offer some insight into X.M.P. and what benefits the idea has.
_______________________________


Test setup:

Asus P5E64 WS Pro, X38 chipset, BIOS 0702.
Intel C2D E8400
Tagan 800W PSU

_______________________________


Pre-test notes, assumptions, drawbacks.

*The board cannot supply less than 1.70V to the RAM, partly through bad BIOS labelling (VDIMM is around 0.1V than BIOS value) and partly due to not utilising the bottom end of the controllers outputs. This is a REAL annoyance.

*The board splits the RAM timings into three groups (primary, secondary and....obscure,) which are either all available for manual setting or all set to auto. The X.M.P. ties into the primary settings, so affects the four main timings we're all used to. It also has a voltage request, but as the board overvolts, this is muted. VDIMM CAN be set manually in BIOS, within the limits explained above.

*The X.M.P. profile does not appear to take NB strap into account which may clash with any expected performance gains at certain FSB areas.

*X38 RAM clocking is closely tied to vNB, as the memory controller is still what typically limits DDR3 clocking. I used assorted voltages values, up to 1.58vNB for testing at certain RAM values but as it was only stock-cooled, I went no higher. This will reflect the real-world limitations for most users anyway. Higher vDIMM values did not overcome vNB limitations.

*Two low-speed 80mm fans were positioned over the RAM at all times

*MaxMem was set to 600MB for all tests. While this strays from a real-world testing scenario, it does help limit test variation as the RAM volume was tuned to the bench test (SuperPi32M.) Because of the nature of RAM (its random!,) using Maxmem was expected to reduce the effect of data placement scatter compared to using the full 2GB.

*Crucials 800MHz X.M.P. profile is rated at 1.90V, as is the standard voltage for the E.P.P. 900MHz kit. I took these to be the maximum warranty values and also tested 0.1V above them to see what the RAM could do when pushed.

*the EPP profiles of the 900MHz kit could not be tested properly as I did nto have a 790I board at my disposal

*All voltages were measured directly using a digital multimeter.
_______________________________

K404
08-12-2008, 03:50 AM
Crucial Ballistix DDR3-1600 X.M.P


Intel X.M.P. offers 800MHz 8-8-8, 1.80V
Crucial X.M.P. offers 800MHz 7-7-7, 1.90V

How X.M.P. shows up to the user is that in BIOS, as well as Manual and Auto settings for the FSB, multi etc, X.M.P. presents itself as an option. The PC12800 RAM comes with two XMP profiles- one from Intel and one from Crucial. Intels profile is fairly tame, but Crucial ramp things up a bit for theirs.

As always, I will not insult Crucials QC or binning by testing at default settings.

First of all, the most logical thing to do would be to bench each Profile at default to see the difference:


XMP profile 1: XMP profile 2:

http://img291.imageshack.us/img291/1197/xmpprofile1an9.th.jpg (http://img291.imageshack.us/my.php?image=xmpprofile1an9.jpg)http://img503.imageshack.us/img503/9748/xmpprofile2tc2.th.jpg (http://img503.imageshack.us/my.php?image=xmpprofile2tc2.jpg)

Not a big difference to be honest- only 6 seconds. Based on those results, I cant see the timings having much of an effect in games either.

____________


Using Intels XMP as a baseline:
From a base of 800MHz 8-8-8, 1.80V.

Moving onto manual settings, I started off with Intels Profile as the guide and tried with lower voltage (1.70V)


http://img503.imageshack.us/img503/3583/800mhz888170vxmpry2.th.jpg (http://img503.imageshack.us/my.php?image=800mhz888170vxmpry2.jpg)

The next step was to try tighter timings (7-7-7) at stated voltage and MHz:


http://img291.imageshack.us/img291/5336/800mhz777180vff0.th.jpg (http://img291.imageshack.us/my.php?image=800mhz777180vff0.jpg)

tighter still: 7-6-6 at 1.80V:


http://img291.imageshack.us/img291/9551/800mhz766180vkj3.th.jpg (http://img291.imageshack.us/my.php?image=800mhz766180vkj3.jpg)

900MHz at the stated timings and voltage was no problem:


http://img206.imageshack.us/img206/5748/900mhz888180vjf8.th.jpg (http://img206.imageshack.us/my.php?image=900mhz888180vjf8.jpg)

So far, so good!




Using Crucials X.M.P. as a baseline:
800MHz 7-7-7, 1.90V
Moving onto the second profile, I went through the same procedure as before.

lower voltage: (1.80V)


http://img206.imageshack.us/img206/2590/800mhz777180vzt4.th.jpg (http://img206.imageshack.us/my.php?image=800mhz777180vzt4.jpg)

As shown before, 7-6-6 latencies at 1.80V were no problem:


http://img291.imageshack.us/img291/9551/800mhz766180vkj3.th.jpg (http://img291.imageshack.us/my.php?image=800mhz766180vkj3.jpg)

some increased MHz (940MHz) at 1.90V and 8-8-8 timings:


http://img291.imageshack.us/img291/3083/940mhz888190vvl2.th.jpg (http://img291.imageshack.us/my.php?image=940mhz888190vvl2.jpg)

so far, nothing to be ashamed of!

The best thing about these results is that they are from retail sticks- not cherry-picked for me. Crucial have a nasty habit of selling their products short! While I appreciate that slacker standards mean binning is less demanding and steady supplies are easier to fulfill, from the Crucial products I have owned or reviewed, I feel confident in saying that they could easily offer higher-spec parts with little fuss.

______________________________


Crucial Ballistix DDR3-1800 E.P.P. 2.0


Moving onto the 900MHz parts, I expected these to easily match the performance of the 800MHz sticks, but as assumption is the mother of all screwups, I started from the beginning. I'll spare the reader the intermediate steps- I eventually ended up with 900MHz 8-7-7, 1.75V!! yes- tighter timings at lower voltage than the official rating. The sticks weren't big fans of CAS 7 over ~ 850MHz, but seeing as not many people will buy a kit like this to run at below-spec speeds, I didnt chase many MHz/timings combos. I would expect these to perform fairly similarly to the 800MHz parts above though.


http://img503.imageshack.us/img503/7721/900mhz877175vnz0.th.jpg (http://img503.imageshack.us/my.php?image=900mhz877175vnz0.jpg)

I did have a quick play at other settings, just to confirm the kit was capable-
Generally it seems to be able to match the 800MHz kits abilities, but with 0.05V less at a given MHz/latencies point

At 1.75V, 800MHz 7-6-6 was no problem:

http://img206.imageshack.us/img206/2144/800mhz766175vhr4.th.jpg (http://img206.imageshack.us/my.php?image=800mhz766175vhr4.jpg)


Sadly, the next option- to increase MHz, was a disappointment. I wasnt expecting much headroom based on the memory controller limitations, but stability tailed off quickly after 910MHz and 920MHz wasnt stable, even for SPi1M. Seeing as the sticks perform very very well at 900MHz, I was surprised that neither slacker timings nor higher VDIMM could lift the limits. Are Crucial protecting the value of their 1GHz parts by limiting the 900MHz kits? Is my motherboard playing up?

I appreciate that the kit isnt really meant for Intel chipsets, so I should be fairly happy I got anything worthwhile out of them at all.

__________________________________

Summing up + Conclusion:

Crucial direct price for these kits are currently as follows:

PC12800: £141
PC14400: £148

For the price difference of these two kits, the performance guarantee of the 900MHz kit makes it hard to refuse, if the mobo they are used in can handle them. However, Crucial are not alone in the performance RAM market...

OCZs DDR3 price war means that they can offer a 2GB set of 900MHz RAM with 8-8-8 timings for less than £95, or an 800MHz kit with 7-6-6 timings for £87. That puts a recommendation in a tough place. Crucial price/spec mix is not ideal at the moment, which forces me to re-iterate a previous comment: Crucial could easily tighten up the timings of their kits or drop the suggested voltage. Until then, these kits, although good, are going to get overlooked for the more aggressive presentation of OCZs products, which sadly goes as far as packaging. Bling blister Vs brown box....

Another point is cross-platform compatability. CSX RAM kits have multiple ratings for all three (currently) major chipsets. This neat touch means users can move between motherboards with an assurance of compatability with the RAM. Most other vendors tune a RAM kit to one chipset, which I can see putting some less well-read computer builders off.

I would like the thank Crucial Europe for the review samples and Joan Lunny for having the patience of a Saint!

______________________
Post-script. Testing didnt stop with those results. I have kept tweaking and a tRAS of 14 did not hinder stability or require extra voltage. Thats an improvement of around 40% over the stock latency! Im not ready to give up on finding raw MHz. Testing will continue and any new developments will be included when ready for presentation.

Oliver
08-14-2008, 12:14 PM
Looks good Kenny, good to see ur are playing ;)...

nut311
09-02-2008, 03:06 PM
do you know thats IC on Ballistix PC3-10600?

MACMAC
09-10-2008, 12:52 AM
do you know thats IC on Ballistix PC3-10600?


I will have non-Ballistix Crucial PC3-10600 before Friday, so I will let you know if you're interested.