![]() |
|
|
#133 |
|
P90 years forever!
Aug 2002
Yeehaw, FL
19·397 Posts |
|
|
|
|
|
|
#134 | |
|
∂2ω=0
Sep 2002
República de California
1164710 Posts |
Quote:
The RAM packaging says "ddr3 2400", so why is the default clocking just 2/3 of that? Is the 2400 the maximum rated speed, but 1600 the 'advisable' speed? [Sorry, n00b to OCing here]. Last fiddled with by ewmayer on 2013-06-19 at 06:15 |
|
|
|
|
|
|
#135 | |
|
"Oliver"
Mar 2005
Germany
11×101 Posts |
Quote:
Oliver |
|
|
|
|
|
|
#136 |
|
"Kieren"
Jul 2011
In My Own Galaxy!
2·3·1,693 Posts |
If it's rated at 2400, you can certainly go there, and it should work at default voltage. My RAM on a previous board was rated at 1600, but defaulted to 1333. I'm not sure why this happens.
EDIT: Now I know. Thanks, Oliver. Last fiddled with by kladner on 2013-06-19 at 11:52 |
|
|
|
|
|
#137 |
|
"Serge"
Mar 2008
Phi(4,2^7658614+1)/2
9,497 Posts |
Motherboards all have different instructions, so the best way to find out is to RTFM of the exact motherboard. I flipped through the ASUS mobo's PDF and it is very detailed and has a screenshot of every step (it is even too detailed).
Also, get the latest BIOS onto a USB stick and reflash if desired. (It probably is not needed for the X87s - they just came out. Or maybe not! Most BIOS patches are released in the very few first months and then there are no updates for years.) Last fiddled with by Batalov on 2013-06-19 at 17:40 Reason: latest BIOS? |
|
|
|
|
|
#138 | |
|
∂2ω=0
Sep 2002
República de California
19×613 Posts |
Quote:
Does the fact that I was able to toggle the profiles in the menu mean that my dimms support xmp? (The above wording is ambiguous in that regard.) I see 'profile 1': ddr3 2400MHz 10-12-12-31 1.650V DRAM Timing Mode [selectable between Auto / Link / Unlink] What does the latter auto/link/unlink setting do? 'profile 2' has same ddr3 numbers ... maybe those are to be able to have one profile 'Auto' and the other, say, 'Link'? Those are the only listed profiles. Will the above xmp settings crank the dram frequency right up to 2400? I'd rather ratchet it up incrementally (say 1800, 2000, etc) and let each setting run for a few hours to check for instability. |
|
|
|
|
|
|
#139 |
|
Feb 2012
6258 Posts |
XMP profiles are there so you do not have to mess with manual settings.
Memory modules may have one or more XMP profiles encoded in them. The idea is, you select the XMP profile you want, save your settings, and that is it. An XMP profile is a shortcut to several manual settings applied in one swoop. You can further tweak individual settings manually, if you want, but that is not required. |
|
|
|
|
|
#140 |
|
"Kieren"
Jul 2011
In My Own Galaxy!
2×3×1,693 Posts |
"Linked" and "Unlinked" sound like what my board calls "Ganged" and "Unganged". On mine, it indicates whether the two memory channels are in tandem or operating independently. I believe that Ganged is possibly beneficial to systems which give most of their resources to a single task. Unganged is more responsive to multi-tasking. (This assumes that the terminology is equivalent.)
|
|
|
|
|
|
#141 |
|
∂2ω=0
Sep 2002
República de California
265778 Posts |
OK, I did a bunch of memory-speed-based timing tests, all using the params of my ongoing multi-month primality test of F28 (known to be composite already via small factor, but full residue still useful), that is Mlucas_AVX running in Fermat-mod transform mode at FFT length 15360K, on all 4 cores of my 3.40 GHz Haswell system. The first 2 lines [test 1-2] are timings using BIOS-set defaults for the 2 kinds of RAM I tried; the next 4 lines [3-6] are result of manually upping the RAM speed from default 1600 MHz to 2400 MHz in 200-MHz increments; the final 3 lines [7-9] are for the XMP-menu preset memory-overclock "Profile 1" for each of the 3 available DRAM timing modes:
Code:
Test# DRAM settings sec/iter [1] ddr3 1333 @1333 MHz: 0.0842 [2] ddr3 2400 @1600 MHz: 0.0755 [3] ddr3 2400 @1800 MHz: 0.0720 [4] ddr3 2400 @2000 MHz: 0.0699 [5] ddr3 2400 @2200 MHz: 0.0652 [6] ddr3 2400 @2400 MHz: 0.0625 [7] ddr3 2400 @2400 MHz, DRAM timing mode = Auto: 0.0625 [8] ddr3 2400 @2400 MHz, DRAM timing mode = Link: 0.0624 [9] ddr3 2400 @2400 MHz, DRAM timing mode = Unlink: 0.0624 I also tried XMP profile 2: 0.0626 sec/iter (i.e. slower, if the difference is even meaningful) ...And disable-turbo-boost: 0.0628 sec/iter, so I re-enabled that. Upshot: Swapping the older 1333@1333MHz for the 2400@1600MHz gives ~10% speedup. Cranking up the clocking on the new memory from 1600 to 2400 MHz gives another ~17% boost. Nice. |
|
|
|
|
|
#142 |
|
Mar 2003
Melbourne
5×103 Posts |
|
|
|
|
|
|
#143 |
|
∂2ω=0
Sep 2002
República de California
19·613 Posts |
Funny thing happened last night - I was in bed but not yet asleep, when I heard the Haswell CPU fan - case is sitting open on my desk about 4 feet away from my head - suddenly slow down for about 1 second, before resuming back to normal speed. That got me wondering, so I listened for another half-hour, same thing happened several more times. Then I noticed that this was happening every 10 minutes, almost to the minute - and my per-iteration time for 4-threaded F28 run is currently 0.0615 seconds. It's the savefile writes, which occur every 10000 iterations ... I use an SSD so there's no disk-write noise, but the brief interval in which multithreaded crunching stops and the current floating-point residue gets converted to bytewise endian-independent form and written to disk is enough to cause an audible "hiccup" in CPU fan speed.
"I can hear it working" (or 'not working', in this case). |
|
|
|
![]() |
| Thread Tools | |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Haswell-E Prelim. Benchmark | sdbardwick | Hardware | 37 | 2015-02-10 18:49 |
| Prime95 and Haswell | Pleco | Information & Answers | 22 | 2014-07-13 16:03 |
| Haswell Rig | Mini-Geek | Hardware | 64 | 2014-05-27 13:22 |
| Prime95 version 27.1 early preview, not-even-close-to-beta release | Prime95 | Software | 126 | 2012-02-09 16:17 |
| Missing mouse-over preview text | retina | Forum Feedback | 1 | 2011-09-12 15:32 |