Rev 'em to 287! -> 2GB Mushkin Redline PC4000 on Infineon CE-5 OCing Report!!!

[bachus_anonym]

Hello and welcome to my next OCing Report!

This time I had a chance to put to test the latest spin of 2GB low latency DDR from Mushkin - PC4000 Redline (991493). Why the latest, you may ask. Well, this time Mushkin started assembling it with Infineon CE-5 components and as far as I was told they have been doing it for several weeks now. Mushkin seems to have gotten hold on a Infineon CE-5 IC week with very good yields, that appears to clock better on average (including 3D) from what we have generally seen so far. I kindly requested a pair that is not cherry-picked, comes from their usual binning, was pre-packaged and not tested after that. I was told it's a random pick but I would still like to stress the fact that, as in OCer's life, mileage and luck may vary when it comes to retail-acquired kits. Nevertheless, Mushkin seems to be confident that this batch is very impressive in terms of OCing potential.

From my understanding, this particular batch has not found it's way to e-tailers just yet. But I have noticed that Newegg has just yesterday sold its last pair of Redline XP4000 so I'm hoping that maybe next shipment will consist of those very sets I test here. I will try to get a confirmation on that as soon as I can.

OK, let's start. Just as usual, time for a photo session.

Sealed package



Heatspreader look from both sides



Side by side



A little close-up



Setup:
AMD Opteron 146
DFI LP Ultra-D R.A02, BIOS 11/14 Official
NForce4 drivers 6.70, USB, Network and Audio enabled.
EVGA 7800GTX 256MB KO @ 490+40/1390 (low-power mode disabled thru BIOS-mod)
Forceware 84.21, High Quality
Windows XP SP2, not tweaked




Part1: MAX *Full stability* at CAS3.0

SPi32M & 3DMark03 @ DDR570 5-2-3-3.0-1T, 285x10=2850Mhz

3DMark2001 & 3DMark06 @ DDR570 5-2-3-3.0-1T, 285x10=2850Mhz

2hrs of SP2004 LargeFFT 1024MB used @ DDR570 5-2-3-3.0-1T, 285x10=2850Mhz

Overly impressive results here and with so little voltage, too! Just for the record, my board overvolts vdimm by 0.04v measured with a multimeter. I have played Far Cry and F.E.A.R at those settings for enough time to proclaim it fully stable. Also, looping 3DMark2001 on LobbyLow test for 2hrs, while lestening to streaming radio (to catch "freezes" easier) was not a problem at all.


Part2: MAX *Full stability* at CAS2.5

SPi32M & 3DMark03 @ DDR540 5-2-3-2.5-1T, 297x10=2970Mhz

3DMark2001 & 3DMark06 @ DDR540 5-2-3-2.5-1T, 297x10=2970Mhz

2hrs of SP2004 LargeFFT 1024MB used @ DDR540 5-2-3-2.5-1T, 297x10=2970Mhz

Maxing out those DIMMs at CAS2.5 required a "very healthy" bump in voltage past 260MHz. But once provided it was enough to gain complete stability at whooping 270MHz! That's something, isn't it?


Part3: MAX *Full stability* at CAS2.0

SPi32M & 3DMark03 @ DDR486 5-2-3-2.0-1T, 292x10=2920Mhz

3DMark2001 & 3DMark06 @ DDR486 5-2-3-2.0-1T, 292x10=2920Mhz

2hrs of SP2004 LargeFFT 1024MB used @ DDR486 5-2-3-2.0-1T, 292x10=2920Mhz

Same 2.9v was required to perform another step-down on CAS Latency. This set did not disappoint here at all. Passing by 240Mhz was as easy walking - just natural


Part4: Absolute MAX at 5-2-3-3.0-1T (not 3D stable)

SPi32M @ DDR590 5-2-3-3.0-1T, 295x10=2950Mhz

2hrs of SP2004 LargeFFT 1024MB used @ DDR580 5-2-3-3.0-1T, 290x10=2900Mhz

Now, this is where 3D tests were too much to handle for Redline. This batch of CE-5 also has a "3D glitch" but it shows up at considerably higher memory frequency. While it's not diminishing just yet, it's heavens better than what earlier samples managed on average

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Update 1: Improved MAX *Full Stability* at 5-2-3-3.0-1T and 1st "divider OC"

SPi32M & 3DMark03 @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

3DMark2001 & 3DM06 @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

SPi32M & 3DMark03 @ DDR570 5-2-3-3.0-1T, 342x10=3420Mhz

3DMark2001 & 3DM06 @ DDR570 5-2-3-3.0-1T, 342x10=3420Mhz


Update 2: Synthetic Benchmarks at 287MHz 5-2-3-3.0-1T

Sandra Mem Band. & Everest Read @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

Everest Read @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

Everest Mem Latency @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

ScienceMark2.0 L1 Band. @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

ScienceMark2.0 L2 Band. @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

ScienceMark2.0 Mem Band. @ DDR574 5-2-3-3.0-1T, 287x10=2870Mhz

[Eldonko]

Awesome work bro, I was wondering what this stuff did @ 2.5, I hadnt had time to test my set yet. No 3d issues = sweeet.

[Cooper]

Excellent work bachus
Finally some real stuff. Those Infineon surely give some competition to Micron and UCCC
3D stable at 285@3-3-2 that`s just awesome

[Revv23]

man, its worth it for that 2.5 speed alone!

we are going to need some testers though, i konw mushkin is reputable, but its quite possible these are better then average.

[Jupiler]

285Mhz, 3-3-2-5 and only 2.50V is quite impressive.
Thx for sharing your results, B.A.

[Deathspawner]

Thanks a bunch for another detailed report! I love it how well those modules thrive on such lower voltage. Great sticks you have there.

[CapFTP]

tuned to another great OC report...thnx BA

[bachus_anonym]

Awesome work bro, I was wondering what this stuff did @ 2.5, I hadnt had time to test my set yet. No 3d issues = sweeet.

I would not say that "3D glitch" is no longer there. In my case, it still exists but this batch seems to show it at much higher frequency, compared to other CE-5/CE-6 we've seen on average.

we are going to need some testers though, i know mushkin is reputable, but its quite possible these are better then average.

I agree... Although, this week appears to be the best so far and there is a good chance many of those new kits will see 275MHz in 3D. Since Newegg has just sold last pair of the previous shipment, I would think that next time it's in stock - it will be with this batch So, I hope we will soon see more of the happy CE-5 OCers

[coop]

Damn nice report bachus. Good rams too. The versitility of different settings should be helpful for many.

[Fayt_Leingod]

Dude! that is so awesome! Coulda been a little better though, i mean we all know VB > Heinekin.

[RafaDel]

Not at Newegg yet... Should I try directly from Mushkin site? Will I get CE-5 from them?

Thanks

[|3ourne]

great review as always bachus ! Love your reviews man , so thorough. I had been talking to mushkin to see if they can send me a review sample but looks like u guys got ahead of me . Ohh well , you guys are better qualified and more knowledgable neways. Keep up the good work bro.

[x3ro]

Nice review dude.. Also nice clocks too!

Regards

[AndreYang]

I think CE-5 is better than CE-6. Good results.

[bachus_anonym]

Thanks guys!

I'm actually still working on pushing it higher in terms of 3D benching/gaming, as we speak. I've 3D-benched it as high as 287MHz 5-2-3-3.0-1T 2.54v (2.58v DMM) but at this speed it doesn't survive even 30 minutes of looping 3DMark2001 LobbyLow... It goes "freeze-and-release" like there is no tommorow So, it looks like 285Mhz is true all-stable MAX for this set - not too shabby

I will post few more benches soon - mainly some "divider action" close to the MAX of my Opteron - it's XtremeSystems after all, right? 2900MHz is for wussies

[EnJoY]

Bachus, what did higher voltages yield?

[@_dud]

I was just waiting for a nice pair of 2x1gb sticks, and this looks like the trick. Thanks Bachus, and nice review.

[bachus_anonym]

I have just refreshed my 1st post with Update 1: 3D benches @ 287MHz 5-2-3-3.0-1T and initial divider action.

More results tommorow

[gigi77]

great result!

[revenant]

nice mems.. been looking for a solid set which will do 280+... sweet!

[high5]

wow, that are some awesome clocks from Mushkin.

i haven't seen much of this stuff, but all the samples i remember were impressive, even the older ones from nugzo and Anandtech.

[bachus_anonym]

i haven't seen much of this stuff, but all the samples i remember were impressive, even the older ones from nugzo and Anandtech.

Damn, AT hit 293Mhz in 3D with their kit, and with CE-6 too? Not that it matters If both revisions are binned in the same way, but damn - that is an awesome result! Nugzo's 290Mhz also got me super excited when I saw it back then. I'm still trying to match his clocks in 3D but it ain't happening
With my kit, more than 2.58vdimm (DMM) does not make any difference at CAS3.0 when it comes to 3D stability. But 2.79v is neccesary to pass SPi32M @ 295Mhz, although 3D gives an instant freeze at that speed.
Also, interesting is that with previous Rev.C kits I tested I could gain extra "32M/SP2004-stable" clocks by loosening latencies to 8-3-4-3.0-1T. This XP4000 Redline I have here, maxes its clocks out at 5-3-2-3.0-1T and loosening does nothing.

I added few more synthetic benchmarks last night and will probably make one more update today.

[high5]

Also, interesting is that with previous Rev.C kits I tested I could gain extra "32M/SP2004-stable" clocks by loosening latencies to 8-3-4-3.0-1T.

how much did you gain?

[bachus_anonym]

how much did you gain?

Not that much, always less 10Mhz but it was at least some gain. This kit doesn't care for it at all...

[Duonger]

hey bachus very nice results... im more interested in the high resolution of these pics. what camera did you use to take it with?

tom