Fatal1ty X99M Killer possible with 128gb ram?

O_and_N wrote: Wardog,thank you for the link.Just took a look at the micro atx they have there and some how im not sure if it is what i want.It says it needs a narrow type CPU cooler and i was planning to put a noctua NH-D15  /  or NH-C14S  on the e5-2690 v4(no water cooling).And i find the Fatal1ty X99M Killer/3.1 to be more feature rich/balanced  for the gaming/game dev part and the rendering part.

Meh, are you sure? Those coolers are a lot of mass(read stress on a motherboard) to be transporting via backpack. Particularly the C14S as its heatpipes unfortunately allow it to unbalance any loading to only two of the four mounting holes. Unduly stressing one side of the cpu socket/motherboard.

http://noctua.at/en/products/cpu-cooler-workstation-server/tdpguide

^^ Excellent chart. I'd look for the shortest (edit) that has heatpipes protruding from both sides of the cpu plate (stop edit) that meets the TDP of whatever Xeon you decide between here and build time.

The shorter the better. I mean, c'mon. That backpack will one day too soon be heavy and a particular PITA that day, only to be nonchalantly flung by its straps over and out of the way. At which point you will gasp aloud and your heart will sink, displacing your intestines.



Edited by wardog - 25 Sep 2016 at 12:59am

parsec wrote: When is a certain speed an "over clock"? When it is beyond the specs for the device.

If device means memory stick I'm wholeheartedly behind you.

wardog wrote: Not on either of you twos lives is 128GB 2133 ECC OC'd to 3000. O_and_N, am I sure your building a production box or a gaming rig? Convince me. One simply doesn't build a production box around overclocking its components. They don't. That 3000 there is most likely an anomaly of production. My money's on they needed to select 3000 binned chips for quality, considering the number of chips on each stick, to run reliably at their rated 2133 speed. I don't believe for a split second they're running 128GB ECC overclocked to 3000. Impossible. Just impossible. Not ECC and it's extra bit. Anomaly. Period.

You could be right about this. I was simply explaining how the Main screen in the UEFI shows the memory speed once it is applied.

Now I see why the OP asked if that memory is over clocked. Not a dumb question.

I would think ASRock would have said in their message to the OP that the speed of 3000 is an OC, or a bug in the UEFI when ECC memory is being used, even with a Xeon processor.

Given these two scenarios, and the lack of an explanation, I must choose a bug in the UEFI.

Or it was able to POST (??) but not boot an OS at that speed.

To play devil's advocate , the 3000 speed shown, if an anomaly, seems rather a random number. Sure there is 3000 speed DDR4 memory, but that is a more unusual speed.

I can't figure where 3000 comes from, not a multiple of 2133, or the actual frequency of 1066MHz for DDR 2133 memory.

Regardless, the 128GB was recognized on a mid-range M-ATX X99 board. That speaks to the quality of the build of that board. In this case, being a smaller X99 M-ATX board with only four memory slots, is an advantage.

Also, the memory controller on that Xeon CPU is capable of handling 1.54TB of DDR4 memory!

Sounds crazy? See for yourself:

http://ark.intel.com/products/91767/Intel-Xeon-Processor-E5-2650-v4-30M-Cache-2_20-GHz

Perhaps that gives it great over clocking mojo, even with ECC memory. So ECC memory has a parity bit, and other features for error correction. Why does that prevent operating the memory at a higher speed? Sure, it's not done for stability reasons, but why is 2133 safe, but say 2400 not? This Xeon CPU is rated for use at memory speeds of 1600, 1866, 2133, and 2400. When is a certain speed an "over clock"? When it is beyond the specs for the device. At one time not long ago, any speed above 1600 was an over clock.

Just sayin'

parsec wrote: The Crucial part number is CT32G4RFD4213.36FA1. We can see 2133 (oddly) included in the part number.

The actual P/N is CT32G4RFD4213 . I suspect from the . period right  is related to production tracking.

http://www.crucial.com/usa/en/ct32g4rfd4213

O_and_N wrote: Note taken hehe. And now for a probably dumb question.In the image above,did they overclocked the ram?

I almost said, did you look at the picture? But if you aren't familiar with an ASRock UEFI on an X99 board, how would you know?

That's the Main screen in the UEFI, which shows, among other things, the DIMM slots populated with... uhm, DIMMs, memory sticks (I hate saying sticks), and the speed that the memory is running at.

We can see it is at 3000. The Crucial part number is CT32G4RFD4213.36FA1. We can see 2133 (oddly) included in the part number. Just to verify, just search on the part number above. One result example:

http://www.serversupply.com/products/part_search/pid_lookup.asp?pid=272137&gclid=CPPohMXNps8CFdgMgQodhkoHFQ

The standard speed of this memory is 2133. I don't think ECC server memory would have an XMP profile, but who knows.

Clearly, the memory is over clocked. It may be manually over clocked, which is no big deal to do.

What is a big deal is 128GB of ECC server memory used with a Xeon E5-2650 v4 (a Broadwell Xeon) processor being over clocked to 3000.

IMO, to expect 128GB of this memory to be able to over clock at all is...

Oh wait, we don't know what the timings are, might not be good enough.